The 32nd ABU Intellectual Property & Legal Committee Meeting in Jakarta (2026) explored how copyright frameworks must evolve to address generative AI, with key insights that human authorship remains essential for copyright protection while AI serves as a tool rather than an author; jurisdictions like Hong Kong and the US differ in their approaches, with Hong Kong's computer-generated works provisions potentially protecting AI outputs while the US requires sufficient human creative contribution, and broadcasters must implement governance frameworks that document human involvement to maintain copyright claims in the digital content economy.
Install our extension to search inside any video instantly.
3rd ABU Intellectual Property & Legal Committee (IPLC) Meeting | Jakarta, Indonesia 2026
Added:Good morning ladies and gentlemen. How is everyone having a good time in Jakarta?
Yes, we are happy to and also we are so proudly welcome you from the ER headquarter conference hall in Jakarta, Indonesia. Welcome to the 32nd Abu Intellectual Property and Legal Comedy IPLC Conference.
Ladies and gentlemen, today's conference will be having the theme AI, copyright and royalties in the digital content economy. But before we begin, please join me in welcoming our distinguished guests. The first one is the chair of RRE supervisory board.
Please welcome Papa Anoir Mujah Adi Tristnanto.
Ladies and gentlemen, please welcome the director of program and production of Radio Republic Indonesia, Papa Mistam.
We also of course would like to welcome the chairman of LMAN, National Collective Management Organization, Baba Marcel Saha.
We would like to welcome the ABU director of news. Please welcome Mr. Indra Singh, chairperson of ABU IBLC, Mr. Haruyuki Ichinoashi.
And of course in this very good morning we would like to welcome all board members of the ABU IPLC honorable speakers delegates and representative of broadcasting organizations from across Asia Pacific.
Ladies and gentlemen to make this morning warmer. We know Jakarta is very warm, but we would like to make it hotter with a very warm welcoming remark from the program and production director of Radio Republic Indonesia. Please welcome Papa Mam.
Good morning.
How are you today?
Distinguished Mr. Indra Singh IBU director of news Mr. Haru Yuki Ishino Hassi, chairperson of the IBU intellectual property and legal committee IPLC head of supervisory board anar moji head of Jakarta Mr. Boogie Hayat Mr. Marcel Shahan.
Oh, chairman of the National Collective Management Organization for Related Rights Honors.
Distinguished members of the IBU intellectual property and legal committee IPLC, honorable speakers, delegates and representative of broadcasting organization from across the Asia Pacific region.
Ladies and gentlemen, welcome to Jakarta.
and welcome to RI.
As the horse broadcaster, RI is deeply honored to welcome all of you to RI headquarters for the 32nd ABU intellectual property and legal committee IPLC conference and meeting.
It is a great privilege for us to host this important gathering of broadcasting professional legal expert and policy makers from across the Asia Pacific region.
This year's conference focuses on the timely and critically important theme AI copyright and royalties in the digital content economy. We fully recognize that the rapid advancement of artificial intelligent and AI is transforming the media industry in unprecedented ways.
Well, AI open remarkable opportunities for innovation and creativity. It also present complex legal and ethical challenges.
This include the copyrights implication content and maybe uh meaning implication of AI training data question surrounding the authorship of genative AI content and editorial accountability within insecuringly algorithmdriven and environment beyond AI related issue.
This committee also faced significant challenges in protecting broadcasting rights, combating digital content piracy and ensuring fair, transparent and sustainable royalty distribution for broadcasting organization and content creators.
In re responding to these challenges, none of us can success alone. We need meaningful dialogues, stronger regional collaboration and greater harmonization of legal and regulatory framework among EBU members to safeguard the future of our broadcasting ecosystem. I would like to express my sincere appreciation to the organizing committee, our distinguished speakers and all delegates who have come together to share their available knowledge, experience and perspective. I am confident that the discussion our over the coming days will distribute significantly to the strengthening strengthening intellectual property protection and legal governance within the broadcasting industry.
I wish you all a productive and successful conference. May this meeting foster meaningful collaboration, generates innovative solution and strengthen the legal framework for the future of broadcasting throughout the Asia Pacific region.
Thank you. Asalam alaikum.
Thank you very much Bameam for such insightful welcoming remarks. And now we would like to invite the ABU director of news Mr. Indra Singh to deliver his welcoming remarks.
Salamat Pagi very good morning and bulah as they say in my uh language. First of all before I start it's an utmost pleasure to be here. Thank you very much Aray for putting together such a beautiful conference for us to be part and parcel of it and always uh great to be in Indonesia uh and you know visit the lovely places apart from the conference etc. So thank you once again to all of you involved in putting this together.
Head of supervisory board for RRI Mr. Anoir Mujay Andy Chrisanto director of program and production Mr. Mist head of RARI Jakata Mr. Boogi Hiday not forgetting the president director of RARI Dr. Ignatius Hendra Mo, all of the RRI staff, the chairperson of the ABU intellectual committee, Harusan, distinguished legal experts, leaders, colleagues from the Asia-Pacific region, and all of those watching online. Ladies and gentlemen, it is my distinct honor to welcome you all to the 32nd ABU Intellectual Property and Legal Committee conference. I was just sitting down and realized 32 years that's a long time and one of the oldest running ABU events that we do host. I stand before you today to deliver this opening remarks on behalf of the Asia-Pacific Broadcasting Union and our Secretary General Mr. Ahmed Nadim who sends his warmest regards and highest expectations for the crucial deliberations taking place over the next day and a half. Our theme this year, AI, copyright and royalties in the digital content economy, resonates very well with public broadcasters, in particular those who continue to face the struggle of fighting and trying to get more revenue through the doors. We are living through a profound paradigm shift. The integration of artificial intelligence into the media landscape is no longer a distant projection. It is here and it is with us right now. From automated newsrooms to AIdriven content to synthetic media generation, the technology is offering access and possibilities never seen before. But with it comes the legal complexities.
As public broadcasters and media professionals, we operate at a critical intersections. We must be innovative, but we must also protect the integrity of our content and of our newsrooms and media houses.
For decades, the bedrock of journalism and content creation has been simple with ethical principles in place and the value returned to the creator. However, this is fast changing. Today, generative AI models are trained on the vast highquality and verified archives of traditional broadcasters. Yet, the legal frameworks that are supposed to govern this data is ingested, how is copyright being recognized and how fair are the royalties that are coming back to media houses who are creating content.
With ABU representing over 220 member organizations, it is also learning of other sister unions like the EBU which recently launched its fax in facts out which has been on the forefront of how we can take the big tech giants to task for using our content. probably a discussion that we can go on and on about. But all this for broadcasters and for media houses is not possible without you. Without the legal brains, without the legal experts, without the legal input, the newsrooms, the content creators, etc. are in the blind because they do not have the expertise. This is where we rely on you.
This is where the legal division of the ABU and those of you present here are the key players for us. The legal committee plays a foundational role in safeguarding our future. Not only the future of individuals or individuals in newsrooms, but for broadcasters.
On behalf of the secretary general and the entire ABU leadership, I want to express my deepest gratitude to the IPLC chair, the committee, the members, Dr. Sema at the ABU Secretariat in Malaysia and everyone involved in ensuring that the content we make on a daily basis is protected. We're not there 100% but probably we could say that we are getting there. Thank you very much for your hard work that you put in. It is time we all got together and provided that shield that protects our creative heritage and the blueprint that ensures our digital future remains ours and not that of third party creators. Once again, thank you very much to RRI. Thank you very much to all our committee members that are present here and I wish all of you a highly productive, insightful and impactful conference.
Thank you very much.
Thank you very much for the welcoming remarks, Mr. Indra Singh. And now, last but not least, let's give a warm welcome to the chairperson of ABU IPLC, Mr. Haruyuki Ichinoashi. The stage is yours.
Okay. Thank you very much and a very good morning to all of you. Um, Mr. Mistam program production director of RRI Mr. Anoir Mujahit Adi Tristan uh Tristanto and Mr. Boogi Hyatt um Mr. Indra Singh ABU director of news distinguished speakers and guests international delegates representing members of the ABU our esteemed RRI colleagues and ABU officials.
First of all, I would like to thank our distinguished RLI colleagues for hosting this valuable opportunity. We very much appreciate your kind and thoughtful cooperation and support.
As the digital technology progresses, our content becomes ever richer and the tech technologies we use to deliver it also diversify both in broadcasting and via the internet. Needless to say, the ways of producing content are improving too.
Conversely, the technological developments and the emergence of AI in particular also raise various legal issues, including in the fields of copyright and ethics.
In such an era, the broadcasters role of serving the public has become more important than ever. As trusted media organizations and content producers, we are expected to provide our audiences with appropriate guidance by delivering accurate, fair, and trustworthy information in the world of deep fakes, misinformation, future bubbles, and the challenges of the attention economy.
In producing and delivering information and content, we also must of course respect people's rights. We can't earn or retain public trust by impairing or impeding people's rights.
Considering all of this, I am sure you agree that awareness of and respect for intellectual property and other legal matters are key for every broadcasting operation including the legal and ethical matters for content production which is this committee's theme.
Our host organization RLI has given deep thought to the theme and topics of this IPLC conference and meeting which is AI copyright royalities and in digital uh economy. It is indeed a timely and wellchosen topic in view of the ongoing trends and issues which face us. I am certain today's conference and tomorrow's meeting will provide invaluable opportunities for us all to learn about and share recent trends, insights, expertise and opinions with each other.
Lastly, and most importantly, I am hoping for two extreme extremely fruitful and productive days ahead.
Thank you very much. Yeah, could you please stay remain on the stage because I I would like to now invite the director of program and production of Eratam to come on to the stage because Mr. Ichinohashi would like to give a souvenir to Pameistam.
Photographers, please go to the front of the stage to to Okay. I would like to invite the photographers to take picture of this moment.
Please everybody. Okay.
Please give big round of applause to Babist and also Mr. Ichinoashi.
And now I would like to also invite Mr. Indra Singh to join us to the stage for a commemorative photo.
Okay.
Okay. Please take the picture in three, two, one.
Okay. Thank you very much. Please make a big round of applause for everyone here and thank you Mr. Mr. Indra Singh and also Mr. Ichinoashi. Please be seated again.
Okay. And now we would like to take a reverse photo here. I would like to invite uh the photographers to go to the stage. And all the participants here, distinguished guests, please give us your best angle to the for our photographers here.
Okay. Should everybody stand up? Okay.
Uh for the everybody please stand up including the ones in the back please stand up and join us in the picture.
Okay. Which is the main camera?
Okay. Okay. In the count of three please. Three 2 1.
One more time perhaps another style. No the same one. Okay. Thank you very much.
Okay. Thank you very much everybody.
Thank you very much everyone.
Okay ladies and gentlemen that is the opening for the conference and please be seated because the first session will begin. So I would like to give the mic to Zuma the host of AU.
>> Good morning everyone. Uh welcome all of you to the 32nd ABU IPLC.
Uh we appreciate uh all international participants and local participants. I understand that this has been a slightly busy morning for some of you as some of you were stuck in traffic. Uh actually the speakers for uh one of the speakers for the first session is on his way. So if we can just wait for five more minutes uh and then we will start with the first session and the moderator for the first session is Miss Manoli Rana Singh. She is the director of copyright for capital Maharaja group a very wellrespected lawyer and an old friend and she will take it from there. Uh but we will just wait for five minutes so that the speaker is here.
Take a break.
Good morning everyone.
I think our missing speaker is here at the moment. So we can start the first session. First session is on AI and copyright framework. And I would like to welcome to the stage our speakers uh of the first session um Mr. Akbad Akmmed Ikbal Tawi. He is the legal analyst, copyright director general of intellectual property, ministry of law of the Republic of Indonesia.
And then we have uh Mr. Desman Chan, Deputy General Manager, legal and international operations, TVB, Hong Kong, China.
And of course, then we have our IPLC chairperson, Mr. Haruyuki Ichino Hashi.
And last but not least, I think we will have Miss Hong Juang Ping joining us online.
She is a PhD candidate of the Singapore Management University.
>> Good morning everyone.
>> Okay, I would like to welcome uh Mr. Akmadbal Tawi uh to take the podium uh for the for your presentation first. Okay.
But okay, thank you very much.
>> Um, his presentation will be on adapting the national copyright legal framework in the era of generative AI and Indonesian perspective.
>> Okay, good morning.
Um first of all I want to say thank you very much for the invitation and but before I begin I'm going to uh say apologies for my director general Mr. Herman Sachar because he can't attend uh this meeting because he has another uh meeting uh and kindly asked me to deliver this address uh on his behalf.
uh for that uh I'm going to uh I'm going to share you about the uh D's perspective about the we call adapting the national corporate legal framework in the air of generative artificial intelligent in India perspective because this is very very unique complex and uh you know it uh disturbing all of kind of uh creation in in in in not in Indonesia but in all the world and that's why we are uh we are now uh make and revision our new copyright uh our uh copyright law uh 2028 2014 and we are revising uh uh and want to to pursue and to the copyright especially dealing with the artificial intelligence next oh Sorry. Uh this one can move.
Yeah, it can. Okay.
Okay.
Wait.
Uh how uh how many minutes that okay is it okay can't move see I'm sorry okay uh so uh we argue that the generative eye disrupt of the intercreative religion uh first that if We uh have a quirks that before uh we uh produce something work that we call it an input before it. We have decoration like a music, books and other uh maybe co maybe uh uh like you know uh other kind of works that are collecting by the data by v scrapping and data set repetition. And the first thing is who gave the permission because the copyright is you know uh it's formed by the declarative system. So we have to we have to uh declare uh uh our our works and also and we have to give permission and the uh the other people that want to use our creation our works we have they have to uh ask our permission to use it and then for the who gives them the permission and and then after that they uh it used uh for the training like uh you know uh like learn the patterns something like that and also So if uh and then output and we cannot make sure is it uh who's the creator and who's the liable I mean the inside is the current legal users not just who owns the output but the the how human works are ingested and utilized within the system. Okay.
Okay. This assessing the three assumption embed in our uh corporate existing law. Uh the first is is it the protected word that can contain a substant I admit the legal requirement of being the part of pre or personal creation of a human because we still pretend that human is a creator. AI is just for the assist we uh is not for the subject of the law.
And the second is the input is the training permitted. Does the user copyrighted works to train AI fall under the existing limitation and exception or does it require a license? And then the the main thing is the who get paid and how catalytic system design for human scale used in uh to identify and compensate creators at the film at velocity driven by AI. And the summary is the exists provide a foundation granting automatic rights to human creators but lack explicit norm for a training data in human contribution threshold and data set transparency.
Next.
Okay. What is it?
Okay. And Indonesia is still aimed to protect human creativity without stifling the innovation, creative protection, defending moral and economic rights and ensuring attribution and fair fair remuneration. But we cannot uh divide that the technology venture also we have to supporting research and education and providing legal certainty for the developers to drive creative industry growth and digital competitiveness. The corporate server block is the human as center. DNS AI is a missionbased system is just as a tool has no legal personality creative field and moral rights. Corporate protects human creativity not pure computation and the goal is to proportional and equitable distribution of benefit risk and responsibilities across the entire creative of value chain.
Okay. uh we have the degree of the human control determine uh eligibility for the protection. First is an AI assisted high human control. Human control the creative conception expressive choice creation editing and final expression is fully protected because just uh me contribution of the human and the second is human AI shared contribution subset a contribution but a human select arrange and revent the output it's by case by case but now we talking about AI generated minimal human control so output lacks sufficient human creative contribution uh let's say we are uh in you know like say the we use uh application like Jimny or cloud or something else we just uh put the prompt simple prompt and then the create journal the create book something else but uh we don't contribute uh more and that creation I think it's not automatically protected because we just uh they use uh fully controlled by the eye so the simple prom is not the main point. We have to make a very very complex problem. We have to ensure that we as a human control the system. We have to uh make a journal something. We we have to take out from the AI. We check again. We make sure this is is it is it is compatible. If you make a regulation with Indonesia regul or nation regulation with contract and make a very very uh you know what I call it uh very very you know uh high contribution in that uh quirks because of that because the output that reproduce substantial part of the previous work potential legal liability for the user. Our prom alone is really enough to establish copyrights. We have to make sure that our prom is very very complex and we have make sure that we contribute more uh to uh uh to create the uh networks.
Okay. And there's more one more the autonomous AI.
It kind of protected by the uh low because it's autonomous they it can think by itself. And we have four step test disting assisted ultra ship from machine output. First is we have to make sure this is the originality for the human is original draft did the human independently compose the prop the prom of creative draft and the second is the editing contribution the human edit range or revine the output and the second is it's the work is uh legally protected by the law we have to make sure is it music book or other creation or other works is it uh related with the science art or literature it depend on the uh the national law of itself and then the distinct character does the work exhibit a personal and distinct character attributely to the human control because is uh corporate is originally produced by human itself it is for character are made the work is classified and as the human contributor is recognized as the legal author okay the trait extend beyond the economic to the identify reputation and authenticity. The first is the omission of the cat attribution itself and the second alteration the work that damage reputation. We have to make sure that the works uh if we uh we can't is it damage rep of the of the um uh the author itself. So we can make sure is that we we if we take or we get the the work like a music or book is it uh fully permitted by the author or not. And the uh the the the problem is that now we facing about the voice and facial cloning or in in US we uh we call it the alvis act the sound the voice the style of the of the alters that's why in our revision we we put them uh as our main substance in I that we have to make sure it's protected or not in low and also the syn synthetic fo and defects They should regulate defense the stent attribution and integrity rights mandating explicit concern for the use of voice or and image and that also the point is mandatory labeling of AI if you if you're using AI you have to make sure you have to to label what kind of AI if you use giminy like this one I'm using notebook LM I'm using the uh the presentation and also the first thing is the establishing to the to the take down, deletion and restoration of the work from the from the outsource and then uh the work used in as fuel cannot remain as a diagrams. I mean the first is we have to make sure the source of the legality was access is underlying works obtained legal and the second is stressibility. We have to has duplicate uh duplication occur within data set and can be documented and also the rever uh reservation of the rights that the right holders have the functional ability to opt out or refuse the usage and the first thing is this is very important for us is commercial remeration that's a commercial use of the training data mandate licensing or not because of the transparency the uh fundamental prerequisite of the all licensing and remmoneration.
Okay. And no no we we're talking about the text and data mining and uh is it one the the main point of our uh revision then uh the first is uh there's any uh limitation from the uh research education and but for the commercial use we have to restrict the transparency of data source. We have to output mechanism and requires individual or collective licensing and demanded fair out uh and auditable remoderation. A single overly brought uh exam 7 TDM would effectively transfer economic value from cats and directly to a developer. I mean uh we have a limitation for the uh using the theat without permission uh education research and also for the uh news or something as uh as long uh make sure they announce uh who is the authors.
Okay. Sorry.
Okay. Now is the leg reverse we can use for uh in our uh regulation that we can use direct licensing if the B2B used for highly additive table of premium catalogs requires direct and perholder license and also we have can you use the collective licensing if it for the mass permission like uh in music uh and the next session is Mr. Marcel my brother will explain to you about but the the must permission because we cannot ask the permission if you are make a concert you have and you have to sing maybe 20 song we have to ask the permission one by one is no wasting time so we use the collective licensing and also with the collective or the uh books and other kind of works that use m permission scrapping distributed through the cmos and the structure of on basic like a music book press visual art and audio visual and and the accountability must follow the proximity of a control resource to risk. The first is you know this is the distribution uh level responsible for handing cooperate report take down at content restriction and responsible transparency or an AI declar de declarations and service security and also responsible for handing corporate reports and then also for the core is responsible for the data set legality system design and algorith algorithmic risk uh mitigation because the uh the principle the end user must not bear the sole burden for the risk that they originate from the flow system design and illegal data set because in Indonesian law if there if there's someone use the work with a permission they the authors can you know sue them in uh criminal or uh in our uh police or we sue them in the court for the uh you know uh uh against the law for using with permission.
So the regulation is reverse is we have to ident identify the the the works we have to make a standardized metad metadata sorry and and universal identifier codes embedded in the content and the second we have to make sure it's the right data and user tracking and the the fourth is payment and uh audit. So make sure this is distributed well and accurate royalty um in music or and uh and um maybe in the books.
Okay. And this one we have to make sure we face road map for adaption adaptation the first guidelines legislation and infra in infrastructure. The first is we have to make immediate action. we have to make uh guidelines in New Zealand. Uh you have uh uh the draft of the regulation for the uh road map of AI is uh I know uh we have a a ministry that related with I from the ministry of uh digital we have the we have they are part of uh AI uh artificial units. So they have to make uh the guidelines and also we have to yeah in initiate multistake dialogues and also not for uh stakeholder we have to like the platform and also and also we have to make a legal revision not only for copyright we also another regulation that related with AI formalize the rules of TDM data set and measureable ops out and the mandate of commercialist licensing and the first uh the last one is ecosystem maturity deploy comprehensive digital corporate infrastructure sector establish auditable AI reality machines mechanism and achieve crossber data interopability with an Asian or global and the five policy pillars from Indonesia a corporate blueprint. The first is the first is center of human protection. The second is mandate of data transparency. The first the third is build commercial licensing using now uh direct or uh collective management system distribute risk proportionality and the strengthened digital in infrastructure. So uh maybe we can uh communicate with uh our Indonesia is the ministry of digital and also because of the future of copyright not choosing between human external and technology. It is ensuring technology works for human creativity, justice and prosperity.
And the second is now we after uh inviting ABU collaboration. The first joint standard developing shared best practice for labeling and attributing a assisted broadcast content and then data interpity the technical uh dialogues discussion and also shared cap capacity uh building joint training between the nation and coordination multilateral voice. So we have to make sure that we have in the same concept same mind to make sure that uh our copyright works and and uh is not you know uh uh protected by law and AI is just as a tools and thank you thank you very much uh and I'll give back to the moderator. Thank you.
Thank you Tyreek.
Next we have Mr. Desman Chan uh speaking on the topic of AI impact on copyright law media law perspective.
Good morning everyone. Um it's my pleasure to join you at the ABU IPLC conference 2026. Now my topic today is um AI impact on the copyright law from the Hong Kong and US perspective.
I promise it's not going to be a heavy law lecture. Uh because many of you are broadcasters or content creators. I will focus on one practical question. If your company spends money creating AI content, can you own it, exploit it, and stop others copying it?
Now, uh let's begin with a short AI video created by TB. As you watch it, please think like a content owner. Uh not only as a viewer, ask yourself who owns this.
Sorry. for a jig pancake.
Fore speech.
Now, okay, in this video, almost everything was generated by AI. The script, actors, music, lyrics, singer, visual scenes and settings. The only exception was the spoken dialogue. TB used real human performers to voice the dialogue because AI could not fully capture the human touch. Now suppose TB spent 1 million US dollars making this AI film five years later someone copies the whole film and puts it on another platform. Can we be stop them? This is the question I want to answer today.
Okay. Now, uh, let me give you the short answer first. In Hong Kong, a pure AI gener is probably protected by copyright because Hong Kong has special rules for computer generated words. In the United States, the answer is very different.
A pure AI generated frame is generally not protected if there's no no sufficient human officer.
So the same frame made with the same AI tool may have different copyright results in different countries. That is why this issue is not just legal theory.
It affects investment, licensing, distribution and enforcement.
For broadcasters, the important questions are who owns the copyright, how long does it last, and will remain protected when the content is distributed around the world.
Now before we answer these questions, let us look at the difference between AI assisted work and AI generated words.
AI assisted work means the human remains the creative brain. AI helps but AI does not take over the creative role. For example, a writer writes a script and uses AI to improve the anguish.
A directive frames a firefighting scene and a teleision uses AI to add frames and exposion effects in post-prouction.
AI helps enhance the scene but the creative work still comes mainly from humans.
Now in this examples AI is a tool. The human is still making the creative decisions.
Put simply human first AI second. The human is the chef and the AI is the kitchen equipment.
AI generated work is different. The human gives instructions but AI creates the expression. For example, the user types, generates a TV drama, writes a film song, creates a virtual actor, or produce an opening sequence. The AI then decides the words, image, melody, look and feel.
So this matters because copyright protest expressions but not ideas.
If the human only gives a broad ideas but AI queries the detail expression then the copyright analysis becomes more difficult.
Now let's look at Hong Kong. Many scholars argue that copyright exist to protect only human creativity.
Therefore they believe AI assisted words may be protected by copyright because humans remains the creators while AI generated words should not be protected because the creative work is done by a machine. Now that will is understandable and it is broadly consistent with the current position in the United States but it is not the position under Hong Kong law.
In fact, Hong Kong's has recognized and protect computer generator words for almost 30 years. This provision were adopt from the UK model and have existed since long before chatb or other modern AI tools appeared.
The interesting point is that a computer generated work is defined as a watch generated by computer in circumstances where there's no human offer.
So in other words, the Hong Kong legislation already comp contempor.
So the Hong Kong law does not simply say no human offer. Therefore, no copyright.
Instead, it asks a different question.
If there's no human author, who made the arrangements necessary for the work to be created?
The law then deems that person to be the author of the work. Now, to keep things simple, I will call that person the avenger.
Think of the avenger as the person who is mainly responsible for making the creation process happen.
This leads to a surprising result. Under Hong Kong law, a pure AI genet work may still enjoy copyright protection. So it would not be accurate to say that only AI assisted words can be protected.
Unlike the US approach, Hong Kong law has had a framework for protecting certain nonhuman creative words for many years. Of course, that does not mean all the problems disappear. The next question is where is the winger? Is it the company that's built the AI model like OpenAI? The company operating the AI platform like uh Microsoft Copilot, the producer writing the palms or the broadcasting or the broadcaster commissioning the work.
Hong Kong law does not yet give a very clear answer on this. The answer will depend on the fact of each case. For Hong Kong broadcasters, the the wis may be less about whether copyright exist and more about who owns it.
The United States takes a very different approach. The starting point there is human officers. US copyright law protest human creativity and US copyright legislation does not contain a computer generated words position similar to Hong Kongs. So in the US if there's no sufficient human offer the work may not be protected at all the leader example is Tyler and permuter Dr. Steven Tyler developed an AI tool that create an image. He submit this image for copyright registration and lease the AI tool as the show over. The application was rejected and Dr. Tyler brought the matter to court. The court of appeals decided that the words produced auton autonomously by AI are not eligible for copyright protection because they lack sufficient human authorship.
Another useful example relates to a comic book called Sara of the Dawn. The story line was written by a human but the images were produced using the AI2 mid journey. The US copyright office allowed protection for the human written test but not for the AI generated images.
Now we can summarize the legal difference in one sentence. Hong Kong asks who is the ranger.
The United States asks where's the human over for broadcast distributing content internationally. That difference matters.
So what does this mean commercially?
It means AI assisted and AI generated words may both be useful but they do not carry exactly the same legal package for copyright subsistence in Hong Kong.
AI assist are protected in the normal way if the usual requirements are satisfied.
AI generator words may also be protected with food to the computer generated words provisions for authorship. AI assist are easier.
The author may be the writer, director, composer, designer or another human creator.
For AI generate words, we need to identify the winger namely the person making the necessary arrangements for moral rights. The difference is important. AI words may carry moral rights such as sorry AI assisted words may carry moral rights such as attribution and integrity rights subject to exceptions and waiver.
Computer generate words generally do not enjoy this moral rights.
For duration, the difference is also important. For literary, dramatic, musical and artistic words such as scripts, lyrics, music and artwork. AI assisted words generally last for the life of the human author plus 50 years.
AI general words last for 50 years from the end of the Canada year in which the work was made.
For films, a normal human create or human assisted film lasts until 50 years after the death of the last survivor among the principal director, screenplay author, dialogue author and composer.
But if a film is purely AI generated and there's no such human contributor, the copyrights last for 50 years from the end of the Canada year in which the film was made.
So the message is not that AI generic words are useless or unprotected in Hong Kong. The better message is this Hong Kong may protect both but the legal instance of difference mobilized duration ownership sanity and international enforcibility.
So what should broadcast do? My suggested best practice is simple. As far as possible, structure projects as human le and AI assisted rather than purely AI generated.
This does not mean AI generated words are worthless in Hong Kong. They they may still be protected, but human le AI assisted words are usually more robust commercially.
for example, longer protection for many underlying works, possible moral rights protection, easier offership and analysis, uh, and a stronger protection in the United States.
Most importantly, check the terms and conditions of AI tools. Use tools that do not take away your copyright or require you to share your rights with them.
And equally important, keep records.
Keep prompt history, prompt revisions, draft scripts, story boards, editing decisions, producer comments, approval records, and vis and version history.
This records help show how much humans contribute to the creation process.
The more human invol involvement there is, the stronger the argument that the work is AI assist rather than purely AI generated.
The same practical advice appears uh or applies in both Hong Kong and the United States.
So let me close with the key message. AI can generate scripts, images, songs, voices, and even dramatic scenes. But copyright law still looks for a legally recognizable corp creator.
Hong Kong is relatively fible because it recognizes computer generated words. The United States is more demanding because it requires human ophip. For broadcasters, the safest strategy is to keep human security in control. Use AI as a powerful assistant and keep good records of the human control.
So in one sentence, use AI broadly but document human creativity carefully.
Thank you very much.
Thank you Desine. Um ladies and gents, I'm sure you have questions uh to ask from our speakers. Um we will be having a Q&A at the end of the session. Now I invite Haruyuki Ichinoashi, IPLC chairperson to speak on Japan copyright office perspective on AI training and copyright use.
Okay, once again very good morning everyone. I wonder if I can use the No.
There you go.
Okay. Actually, I moved the slide too much. Thank you.
Okay. So, today I'm going to talk about the use of AI and related issues in the field of copyright from Japanese perspectives. My presentation actually u may duplicate with the other speakers uh well presentations but it probably tells you that well that kind of topic is sort of universally common and if you find other aspects sort of unique maybe it could be a kind of a locally unique I suppose. So okay so AI technology as you know is progressing quickly and use of AI is also spreading out rapidly as we are aware there are both merits and possible issues to the use of AI. Uh for example AI can improve uh productivity and work efficiency. On the other hand, the issues include the uses of confidential information and personal information and of course you know copyright infringement can also be an issue.
So now we see the three phases of AI on this slide.
First AI model developers develop and train AI models and copyrighted works may be used in this training. This is called the training phase. Second, users use the AI model provided by the AI provider to produce output. This is the utilization phase.
And third, as a result of this utilization, generated generative AI is created as an output. This is the output phase. And let's see um look at the look at this phase by phase.
Okay, first and the training phase.
Here the copyrighted words may be used to train AI models.
I must remind you of the core premise that the user needs to obtain authorization from the copyright holder in order to use a copyrighted work. This is so-called copyright clearance. In this regard, the initial question is whether or not copyrighted words are being used for AI model training. If yes, copyright clearance arises in principle. However, some jurisdictions may provide limitations and exceptions that can be applicable to the use of copyrighted works in AI model training.
And if such a limitation and exception exists and if the way a particular case of AI model training uses a copyrighted work corresponds to the criteria for that limitation exception, no copyright clearance is necessary.
Also as a matter of choice for the copyright holder, some jurisdictions may also provide an opt out that allows the copyright holder to exclude a copyrighted work.
Okay. Second, the utilization phase.
Here the user is using the AI model and the output will be created. The question here is the mere action of using AI to produce an output can constitute a copyright infringement.
How about using an output which is so-called a generative AI? Also, are there any limitation exceptions that can be applied here to the use of AI and its output?
And third, we have the popular copyrightability of the output. The question here is that see as the output is created through use of AI model. Can this output itself be a copyrighted work? Is it human created or is it an output a work which expresses or employs human emotions or creativity.
I understand these issues are under discussion in a number of countries and in Japan a committee appointed by Japan's agency for cultural affairs has discussed and published a general understanding regarding these questions.
The committee noted that its conclusions represent just one way of thinking and are subject to change in the future.
Also that they are not legally binding or do not constitute specific legal opinions. So please keep these things in mind regarding the training phase. The copyright law of Japan provides an exception for the use of copyrighted work in the case where the work is not used for enjoyment of the thoughts or sentiments expressed in it. This is termed exploitation of non-enjoyment purposes and includes both AI development and other forms of data analysis.
In general, a user needs to obtain authorization from the copyright holder to use a copyrighted work. But when this exception is applied, the copyrighted work can be used without the copyright holders authorization.
However, this exception is not applicable in cases where enjoyment is the purpose or even when it is not the main purpose but enjoyment is involved to extend.
Also, there are certain other conditions. For example, it doesn't apply in cases where use would unreasonably prejudice the interests of the author or copyright holders.
Second, the utilization phase. The committee's general understanding states if an AI generated image or any other creation is found to have a similarity with dependence on an existing copyrighted work and no copyright exception applies, it shall be considered as a copyright infringement.
This principle is the same as for existing copyrighted works and there are no new interpretations.
So the question of the copyright does again depend on how the AI model is used and how the output is created and used in the Japanese copyright framework.
Similarity and dependence will be a key to consider whether an output needs a copyright clearance as a copyrighted work.
And third, output and copyrightability.
Under Japan's copyright framework, the term copyrighted work refers to a creatively produced expression of thoughts or sentiments that force within the literacy, a literally academic, artistic or musical domain. On this basis, the committee's general understanding states material autonomously generated by AI will not be considered as a copyrighted work.
Conversely, if a person uses AI as a tool to express thoughts or sentiments creatively, that material will be considered as a copyrighted work and the AI user as the author as a copyright holder.
So you can keep this things in mind and when you use the copyrighted work in Japan and of course this is the only case of Japan. So for example uh this one introduced in the cases of Hong Kong and United States and they are different. So uh the discussions and legislations are subject to individual countries and in addition to these three phases there are some other perspectives now under discussion and sorry I missed it but anyway uh the topic can be for example transparency accountability and traceability will be a keys as well and now uh I am going to talk about some possible measures is uh for mitigating risks attached to the use of AI.
As I explained, uh AI model developers could be using copyrighted works when training AI model. Also, copyright clearances are necessary except when any specific exceptions or limitations are applicable to the use of copyrighted works for AI model training.
Keep these things in mind. What measures can mitigate the risks of copyright infringements and what ought an AI user bear in mind when using an AI service?
And first, when using the AI model, the user should assess the risks by checking how the AI model developer used copyright works AI model development and okay. And second, um, when using the AI model, the user may mitigate the risks by learning how the AI system provider provides its AI service. The points are, for example, whether the AI system providers used any technical measures to prevent intellectual property infringements and whether the AI providers have any policies in place with regard to copyright infringement, etc. And keeping these things in mind, there are several perspectives that a user like us broadcasters ought to consider when using an AI model or output.
For example, as I explained, has the AI model developer obtained copyright clearances for AI model training? If not, have copyrighted works been used for AI model training within uh within the framework of limitations and exceptions.
Are there any technical measures to prevent infringements?
The question here is whether the AI model that you utilize makes proper use of copyrighted works and in addition to it does AI provider disclose information about the database or works it used in AI training? Does the AI provider make its terms of services public? And as I told you, transparency and accountability are also important as well.
And these are the other matters on the policy of each AI service provider. And now how about the policy of the user?
The first question here is how has an AI output been created and used by uh used by the user. For example, does the users organization provide staff training and lay down rules for AI use? Proper training and uses that observe the rules can mitigate the risks that end in copyright infringements.
Also, does the use of AI output fall within a framework or limitations and exceptions or does it require copy querances for that use? It is obviously safer to consider this point before the output is used by the users.
And what about other possibility of copyright infringements? For example, is an AI output similar to any existing copyrighted works? Users can check other copyrighted works on the internet for example before they use it and users al are also advised to record the prompt they have used for the purpose of traceability.
These are the perspectives that can mitigate copyright infringement risks when users like us select and use an AI model and then carry on to use an AI output.
So this concludes my presentation on copyright perspectives relating to the use of AI. These discussions on use of AI are taking place all over the world and I have endeavored here to mention some of the some of the issues.
Well anyway uh please ignore the slide now because all the slide I have know the all the slides. So uh but anyway with these things in mind there are aspects that are not clear yet or may be subjected to change and in many cases interpretation looks like progressing on a casebyase basis depending on the how AI model is used and how the output is created and used. In any case, it is certain that we do need to comply with respect the copyright frameworks. When we use copyrighted works to create and use content, we do need to respect compliance and content production and exploitation can't be cut off from the copyright frameworks. This remains the same in using AI too. Thank you very much.
Thank you Haru.
Now uh we move on to our final speaker who will be joining us online.
>> Hello. Can you hear me now?
>> Yes sir.
Yes we can hear you.
>> Okay. Is slides there? Yeah.
>> Yes.
>> All right.
Good morning everyone. Greetings from Singapore. I wish I could be with you in person, but I'm also very pleased to join you online today. Given the topic of this session, perhaps appearing digitally is quite appropriate.
Today I will speak about generative AI output, copyright ability and authorship challenges.
The corporate debate around the generative AI has a two distinct sides.
The input sides concerns the use of copyrighted works for AI training. This include remination as well as copyright exceptions such as fair use and text and data mining. The output side writes a different set of questions. Is a generated output copyrightable?
Who may be cons recognized as its author and owner and does the output infringe copyright in an earlier work. These questions are related but legally distinct. My presentation today will focus on output side particularly copyrightability and authorship.
Why does this matter? Here are two examples. In 2017, Shia eyes published a Chinese language poetry collection and in next year in 2018 air generator portrait was sold at at auction for more than $400,000 US. Both examples actually came before CHBT and show that AI was already being used in the creative industries and generating commercial value. As AI generated content becomes more widely used, individuals and organizations need to know whether it is protected by copyright, who can claim the rights and how they can prove their claim. These questions begin with authorship because authorship helps determine whether copyright exists and if it does, who initially owns the right?
The pre legal position is that an author must be human. Some scholars have proposed a regularized AI authorship particularly for purpose of attribution and transparency.
But this remains largely a theoretical proposal and would require significant legislative reform.
So under current law the practical question is usually not whether the AI system can be the author but whether the human user was contributed enough to qualify as the author.
The first example is a Jason Allen case sought to register an image he had created a user meter journey. He stated that he had entered at least more than at least 624 times before arriving at the initial image. He then edited it and enhanced the image. The US copyright office asked him to exclude the AI generator element from his copyright claim. He declined and then the registration was refused.
In 20 in 2023 the review board affirmed that the refusal and finding the middle journey generated element lacked human authorship.
So the lesson from Alen case is not that using AI prevent corporate protection.
Uh in 2025 the corporate office report makes several key points. First, the use of AI tool to assist rather than stand in for human creativity does not affect the availability of corporate protection.
Second, corporates protect the original expression in a work created by a human author even if the work also include AI generated material. Third, corporate does not extend to purely AI generated material. as a speaker mentioned before this is different from Hong Kong uh or material whether there is insufficient human control over the expressive elements.
Moreover, the report emphasized that whether human contribution to AI generous output are sufficient to constitute authorship must be analyzed by on a case-byase basis. It also state that based on current uh available technology prompt do not allow provide sufficient control. At the same time copyright may protect the human expression that remains perceptible in the output. It may also protect the creative selection coordination or arrangement of a material as well as creative modification made to an AI generates output. So the key question is not simply how much work the user performed but whether identifiable human authored expression can be find in the resulting work.
Um yeah we just now also heard about this case there are some overlap. The terror case give a clear example of human authorship requirements like Allan who claimed uh unlike Alan who claimed that his own extensive involvement support authorship. Tyler actually deliberately identified the II system itself as a sor author and himself as a copyright owner. The US copyright office refused registration because the work lacked human authorship. The district court upheld that refusal describing human authorship as a bad rock requirement of copyright. The DC circuit later affirmed holding that copyrightable works must be authored in the first instance by a human being. The key takeaway that under current US corporate law an II system cannot itself be recognized as the author.
Turning to China the approach is also human centered but the assessment of human contribution could be uh slightly different. This is an example new in that case Mr. Lee used a stable diffusion to create an image and posted on the social media and Miss Leu later used the image without permission and removed this watermark. The Beijing interled court find that the image felt within the field of art and was expressed in a certain form. It then exam the orange ality and international contribution and ultimately held that the image qualified for corporate protection.
The court find that the least selection and arrangement of prompt parameter settings revisions and the final image selection and reflect his international investment aesthetic choices and personal judgment. It therefore treats a stable diffusion as a creative tool rather than an author and recognize Lee as a human author of this image.
Uh actually this case was attract crit criticism from some scholars and the reading remains uncontroversial but this decision does not mean that every air generates output will be automatically protected in Chinese law.
Uh here is another example phone uh v dong. This is uh in this case m used a middle journey to generate a series of images showing transparent chairs shaped like a butterfly wings. She later alerts that the designs had to be copied.
However, boom had not provided sufficient record to show her credible contribution to the generation process.
She also acknowledged that because medjist output involve an element of randomness. So the same prompt could not produce the same image. Eventually the court held that he had failed to establish orange law human authorship and that a gener image did not qualify for copyright protection. So contrast it's not simply copyright protection in leaving new and no copyright protection in form. The deeper difference whether the climate could demonstrate or prove the sufficiently identifiable human contribution.
Let's now turn to Singapore. Although in Singapore there's no reported the case that uh directly address copyright protection for generative AI output the emergent position can be seen in existing doctrine public guidance and a recent reform report. This including the 2024 IPOS SMU landscape report uh which I also co-author and also the IPOS 2025 public guidance on AI generates on content and the Singapore Academy of Law 2026 report.
So Singapore's current position can be summarized in three points. First and a authoral work must have a human author.
Second, there must be sufficient engagement of the human intellect that uh engagement must be sufficiently connected to the expressive element of the output. Third, significant creative or editorial control may help demonstrate that connection. So where the collection is sufficient remains a question of fact and degree.
This bring us to a central unresolved question. How much and what kind of a human contribution is sufficient to establish authorship? At present there's no fixed or uniform standard. Actually currently I'm writing a paper on this uh on this topic. Um so the assessment remains case specific.
The uncertainty also shows why the familar distinction between AI assisted and AI generated works may be too simplistic. In practice, human and AI contribution are often entwined and the degree of AI involvement may vary at different stage of the creative process.
As a result, the line between the two can become increasingly blurred. Rather than focus on output into um one of two categories, it may be more useful to view human contribution along a spectrum or degree. The law should therefore look beyond the labels and examine the actual creative process. The key issue is not the label itself but whether there is sufficient human author expression in the final output.
Having compared the United States, China and Singapore, we can see both convergence and divergence. All three rem jurisdiction remain human- centered and do not recognize AI itself as the author. They differ however in how they assess whether human contribution is sufficient to establish authorship.
Because AI generated content circulate across bottles, this difference creates legal uncertainty. Greater international coordination and where possible greater converia used to assess human contribution would improve legal pred predictability.
So what does this means in practice?
Document the human contribution. First preserve the input and the process. save the prompt platform and the model details and the relevant settings date and so on. Second, record creative choice and edits. So you can make a note of particular output selected, rejected or revised or combined what specific human change were made. Third, retain both original AI output and the final version. The M's not simply to show how much time or efforts were spent. It's also establish a clear and traceable connection between the human creative contribution and the final work. Of course, documentation does not guarantee copyright protection, but without it, proving human authorship may become considerably more difficult.
To conclude, there are three key takeaways. First, the use of AI does not by itself prevent copyright protection.
Second, AI systems are not currently regulized as authors. So, third, the difficult question is not whether there was human involvement, but whether the human contribution was sufficient to establish authorship.
Thank you very much for your attention and I think the Q&A part will be the most human part for this presentation. I look forward to our discussion.
All right, ladies and gents. Now we move to the Q&A section. Um, do we have questions from the audience?
I'm sure our speakers would be happy to answer.
Any questions for our speakers?
Okay. Uh, I'll start I guess. Um, Tafi, you went first so I'll start with you. Uh, okay. So we have to agree that legal fraternity is basically playing catch-up at the moment because uh technological technology is moving at a much faster pace. And we also have to accept that uh most of the gen AI uh systems that the broadcasters or any of the users are using right now uh they are probably trained on content uh without authorization material obtained without authorization. So uh can you tell me in Indonesia whether the CMO infrastructure um is evolving at a pace to address this? How are you looking at this?
>> Okay, thank you very much. Actually the CMO Indonesia is you know uh not only not only I mean uh in music uh not for the CM other uh kind of works. Yeah. I mean uh the CMO Indonesia is just related with music first and then we are going to uh have another CMOS in other we have you know uh CMOS for books and literation we call PRCI and related with Ira also uh and but it's still not uh I mean can't take the you know how to collect and because we have to uh they have to build the the mechanism uh uh the uh the mechanism royalty the mechanism how can they uh you know take the royalty from the platform or because it it's not simple. So uh that's why in our revision of uh corporate bill we make sure the first is we have to uh protect the the works itself the first the second we have to make sure the ecosystem the mechanism is comply with international rules also and not only for music we have to books and for the uh for the artist reserate also and is the related with AI they make sure that the uh the compensation is paid well but we have to make sure is it uh because in in our existing law is not really compromised with that I I mean it's not uh built in uh because it's you know uh it's from 2014 it's long time ago it's not dealing with the AI so we have to revise that and is the it is the initiative or people uh of our DPR so we uh but we have make sure that uh the the substance of the platform the mechanism of the royalty and like the my my uh my PPS so uh because not really simple it's complex we have make sure like uh broadcasting just like journalistic so we have to make sure that uh all the works all the news or some uh they uh they use for the train uh for the training that they have to pay we have to permit to have the permission of from the authors So uh it's a it's a hard work from from Indonesia to uh to have this uh uh situation. So we must make sure that our revision is uh yeah uh the uh to to strengthen the the the substance of the how we are dealing with the platform like Google like you know like uh let other kind of perform like music also. So we have to make sure that uh uh I mean uh we cannot uh make a same uh same mechanism with music because the journalistic is different with music and from the books also but we have to make sure it's uh in our uh copyright bills uh the first is that everyone that use the c the the work from other from the authors have to uh uh have to gain the uh have to permit if they if they use it with a permission it's illegal that's the first uh they have to make sure and then we have the mechanism like music for the collective management and maybe for the platform it depend if if is related music also for using uh CMO or maybe for journalistic maybe we can uh we are still in discussion is it uh we can use both system is it direct licensing or maybe our collective management system is depend on the agreement the is If there is a B2B it should be direct I mean from the platform and the news uh publishers uh like that and we still discussion with our DPR side so which system that we can use but for the music uh we already use for the collective management system is all extended collective lesson so we collect all the music and then we give uh and we distribute uh the pen but in our uh revision we have make sure it's based on the data. We have to make sure if the uh method that is correct or not like uh music is who's the writer of the song or who's uh who is the arranger who playing the guitar who playing the piano or something we have to make sure that correct that why uh beside we we we we make sure the mechanism we have to thread our our uh our data in it in music we built system we call PDLM national music uh uh central data. So like uh we have to make sure the song uh uh it uh they put the data as uh uh as well as we we mandate in our regulation uh and so the ISRCW and the code is uh we have the code also and also we have to and uh and then we make sure the books also and from other work kind of work.
uh so uh that's why we the main point we revise our regulation is we have to uh we have dealing with the AI and not not also for the eye also for the you know for the other work that not really fully covered in our existing law so we have to uh we have to still uh learning the mechanism so uh like journalistic we have to we can uh we want to discuss with other with German with Dermark how we're dealing with the CMOS in journalistic and also uh and we are we are going to discuss with IRO how to dealing with the CMOS in me and in books and literation and and also we are uh couple months ago I I attended the international conference of res uh artist res right in Philippine so we are going to uh uh learn how to uh dealing with the CMOS in uh you know uh reser rights. So I think uh we are still uh in the position.
>> Okay, understood. Uh Jens, if I were to ask the same question from you. Um Desmond, you might have an opinion on that. Uh how does it work in Hong Kong uh or any other territory that you are working for? Um how's the infrastructure being uh developed or how is it evolving to address this?
Okay. Now, first, um, do you know why there's no question from the floor? Because they're using AI to answer the questions. Yeah, that's that's the experience that I got from the university because I also teach part-time uh at the universities and when I deliver the lecture the students always noted not their heads they typically caught my lecture and then the the a they use AIS to generate a not for themselves and they ask questions instead of asking me they ask the AI to answer the questions.
So that's the day-to-day routine that has become the day-to-day routine. Okay, coming back to your questions now. Uh in terms of the compensation for the uh creators in Hong Kong, is this true that AI is still something very advanced and uh the law is always lacking behind the technology? So we don't have uh at the moment we don't have a very sophisticated legal framework to deal with the compensation issues for AI gener assist and uh my quick answer to you like what I deliver at my presentation is that we always try to argue that well AI is just a tool. It is like uh a pen in my hand or a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a computer on my desk. Okay, I'm just using it as a tool to create the work. I'm still the mastermind of the creation.
So the the traditional applies to my work despite I have used AI to assist me in creating the worst. In this way, I under the traditional law may still be able to came the adequate compensation uh for my creation. So because there's gray area um then there's room of arguments. Uh so we always tried to protect our position by sticking to what has been existing in the current law.
All right? And that's why the important message to you today is that well don't tell anybody that well this is my AI generated work. Don't say that it is my AI assistant work. Okay. I am still the mastermind of the of this piece of work.
So that's the quick answer.
>> Right?
>> If this is not enough then please uh type in the question to Microsoft copilot and get a more detailed answer.
>> Right. Um if you were to advise a broadcaster about what their plan for the next three years or five years um what would you ask them to prioritize uh with regards to this matter?
>> Okay. Um I think we should do it now rather than um later. Uh as my company is also very uh advanced in using the AI technology. We have been using different models of AI technologies in the process of our creation. For example, the demo that we showed to you, it was one of the very first kind of AI productions that uh we broadcast uh in Hong Kong and in our overseas jurisdictions.
So my advice to you is like I said, document your creation your creation process. Okay? Just like my fellow speaker uh mentioned to you in her slides that you need to keep the initial prompt and the final output. And not only that also the um the revision that you have been you have made in the midway of the creation process because this is a very good piece of evidence to show that well we human beings are still uh controlling the whole creation process. We talk to the AI. We don't let the AI dictate us. Okay? Instead, we dictate the AI. We talk to the AI and we give direction to the AI. We give instruction to the AI how to create the word or how to revise the word and how to improve the work. Okay. So, uh keep the records. It is very important. And even more important is before you use the AI tool always with read the TNC of the of the AI tools because we have experienced that some AI tools in the small pins they will say that they will share the copyright uh uh with us or some of them even say that they they will get the copyright of the output of the work that's uh created by the AI tool and for this AI tools do not use them all right because you don't want to get yourself involved in any legal dispute whether the final output uh is wholly owned by you in terms of the copyright ownership. Okay. So the uh there are quite a lot of AI tools uh which do not take away your or share your copyright ownership and don't wait the TSC only once uh because the TSC could be revised by the by the platform.
Uh so you may need your legal staff to regularly revisit these TCS to make sure that um your corporate ownership could not be challenged uh in the subsequent years.
>> Thank you. Um Haru, you work for a public broadcaster.
Um could you share your opinion on uh whether public broadcast public broadcasters given the given their mandate on people's interests putting that forward first whether the stance on AI for public broadcasters and private broadcasters or commercial broadcasters be different?
Well, according to the types of broadcasters, well, it's I think it's up to individual countries, I imagine. But, uh, when I think about the Japanese system, I would say basically uh there are no differences about use of AI. Well, for example, I work for NHK, which is public broadcaster. So, uh, how can I say?
Well, people watching our activities much more strictly than commercial broadcasters I must say. So, uh we are sort of discreet about any kind of activities but uh on the other hand uh AI technology is progressing of course and uh there are good size aspects as well. for example, it can make the work you know go more smoothly and um uh the work efficiency will improve as well. So it depends on how we use it anyway. But uh when we as we are broadcasters we also need to think about you know how our activities will give an impact to the people particularly for example copyright owners including performers for example. uh you might know that uh many copyright owners are beginning to address the concerns about you know the use of their copyrighted works uh for AI issues and uh Japan is not an exception either. So we always need to keep a balance you know between uh how we can properly use the AI and how we can also care about you know the and the respect for other you know copyright holders uh copyright owners rights as well. I think that is comes to my mind about your answer about your question.
>> Um okay uh do we have questions from the audience? Yeah. Uh would it be possible to hand over mic to >> Hi, this is Estra from Turkish Radio Television Organization Cooperation in Turkey. Thanks for all uh for your uh detailed information about your legislation. uh these are really beneficial for us and I would like to ask uh to Hong Kong aspect because this is really how can I say difference for me and for my uh country uh it sounds different in general legal perspective because of you know that in general we are we are seeking human creation but you mentioned uh you said that for example AI generated work sometimes you don't need to uh find uh human creation for this. It is really how can I say different for in general aspects of the legal perspectives on AI. So I I would like to ask the main reason maybe because uh why uh your country how can I say to this perspective to AI? I guess I found this really encouraged to technological developments. So uh I would like to ask you thank you so much.
Okay. If I understand the question correctly, you are asking well if it is AI genital work and there's no human authorship then how can how how can you claim there uh copyright protection?
>> Sorry I can't hear you.
>> I will I think she's asking >> the main reason under this perspective and maybe she wants to know.
>> Oh, okay. I see. Sorry. Sorry about that. F is the explanation. Okay. So how come Hong Kong has a computer generated progressions to protect AI generative work? And uh the answer is very simple.
The Hong Kong government was very lazy.
It simply copied the UK legislation which is this CPDA the copyright protection and uh the copyright patent and design yet 1988.
Hong Kong has this piece of copyright ordinance and added in the 1997 almost nine years um uh after the UK promuggate it legislation and there was a copyright generated provision in the UK legislation and the Hong Kong government simply copy almost 99% of the entire uh uh provisions of this into the Hong Kong law and uh would would say that well they were very smart in contemplating that 30 or 40 years later uh there will be AI generation in the world. Of course not. So it was purely accidental that uh they make that kind of provision in the law and the law could be interpreted in in in such a fible way that it covers the AI technology nowadays.
>> Yeah.
>> I hope I I have answer question. Yeah.
Thank you so much.
>> Yeah, I would like to add on this one because I happened to uh Hello. Yeah, because I happen to do a research on this. I can add a little bit on this part.
>> Can you hear me?
>> Yes, we can.
>> Okay.
>> So, uh yeah, it's actually currently there is a big I think uh we don't have a disagreement on like AI itself cannot be author. So the key issue is uh when the human can be considered as author and then uh I closely examined like two case one is I also mentioned in my slides one in US the Jason Ellen case which give more than 600 prompt and the leu case which is get copyright protection for that AI generated image if I don't remember wrong it's more than 100 times prompt however the judge in China think that uh we do not need to considering about which part is AI generated which part is a human we considering the whole process and the human make the personalized the judgment and the human expression eventually demonstrated by the whole image and also the judge particularly said that we wanted to promote um human to use AI to create that's one of the key reasons We give the protection of u it's not it's not say that purely I generated it's a human use AI as a tool to generate however the US will say that hey I do not want you uh to register the AI generate part you need to tell me which part is human which part is AI which is for Jason Ellen case you you find out Jason Allen refused which is very hard in some way to to say this image you human and AI together to eventually make this image which part is human which part is AI it's very hard so uh so right now there is a sort of uh different judicial practice >> and also in the uh academic there is also a debate some people strongly support yeah let's be author even so my myself I'm also uh currently as I said I'm writing a paper on it and I think I think the the some judges said that we want uh uh promote using AI then it's promote human creativity but some people will say if you give corporate protection to to like a leave case that image they will damage human creativity so it's really what side are you on so depend on yeah >> thank you Sarah >> okay uh >> uh if you very quickly move >> I have a question uh thank you my name is Harrisman I'm uh I work for public broadcasting in Indonesia. I work for TVRI for 26 years now. And as you know, TVRI was established in 1962.
And my question is uh maybe is also the question for other public broadcasters.
How can we protect our uh content? We have archives and also original content since 1962. How do we protect our content and our archives from being uh used by AI models without authorization or maybe fair compensation?
>> That's that's my question. I think >> for training purposes. Yes, imagine.
Yes.
>> How can we protect our archives without being used by AI models?
>> Right. uh would any of the speakers >> like to comment on that?
>> Okay. Uh the first they have to uh from that the corporate is legally uh protected by uh in declarative system.
So uh the since they published is already protected by the law. And then we have uh and the the second is uh you have to make a second step. you have to make a recordation of your uh your corporate your content. So we have uh legal data in Indonesia. So uh that's why in our revision we have to uh we we make to build like in journalistic system uh mechanism of the royalties in in journalistic that related to platform AI. So we have to make sure that the data is uh the data that they take to as a training data is uh they can uh uh in uh transparent and also they have to make sure it is uh uh content is legal or not. So but how we can pro how the how Indonesia can protect the the content if we we don't have data. So we ask uh yeah I mean uh the journalist or the publisher we have to to make a recordation in Indonesian system. So we have so so we have data but if it's too uh if there's mass mass data if there a lot of contents so uh we have to make sure that you have uh you know uh documentation the archive documentation is belong to you uh the first time you publish is already protectable but uh but when the when the uh the platform AI uh the I like jetpit uh jiminy or maybe clo take That's that's a point that uh in our division we have to make sure it's or it's uh uh is protected and make sure the the mechanism was of royalty is uh is uh fully uh you know controlled by the government like like music uh they have struggling with that also we have a lot of music and then uh they using and I know some that they have to uh in our religions is they have to uh you know we have to have a permission from the content creators, from the authors or from the from the you know like your uh your company as a right holder. But uh we still in discussion in the what the mechanism with the with the exact mechanism that we that we can use. But uh the main point is if you uh feel that your content uh is being abused or being used and you can prove uh by the law. So you can use the uh mechanism of the of the criminal uh uh mechanism or we actually we have uh a mechanism of the you know uh uh uh what not this and take down in our uh in our ministry. So we can uh you can uh give us uh like uh what kind of platform what kind of you know uh you know uh like that uh people or content that may be used by other people and then we can block them in the system and or we just uh in ministry of of law is just give the recommendation but the execution is in ministry of uh digital comedy. So we so uh uh you can uh you can contact them also because they have a uh you know a you need that concern in AI the the uh we call it in uh directorate of AI. So you can uh we can uh talk to them if in in AI site. So but in in our regulation if you have have a proof that your content is being abused or used by the AI for train data and you have proof so you can uh you can come to us and we can uh make sure that we can give them uh give the minister of committee recommendation to block the the site but the important is that we can we just can block uh the you know the content is in Indonesian area so we make sure that is in this forum that can we make a you know MOU what or maybe some kind of agreement that uh once it's blocked Indonesia it also block in Asian so they can uh they can use they can uh open the site or in Indonesia so the problem is the weakness is in our regulation is just block from Indonesian site in this regulation but if you use VPN if you use just one step ahead you to go to Malaysia you can open the sites so it's the problem So in this forum yeah I I want to ask you can we have him an MOU or some kind. So if we uh block the system block the site Indonesia it's also give a notice from the other country to block also the system or the the content or maybe the the platform or some uh or the any the sites. So make sure that it's it's it's block from uh in regional uh way. So I think it's the next step. So I think it's just for me.
Sorry I will be very brief. Uh I probably in addition to that we probably need to cons uh take into account the nature as a broadcasters as well because broadcasters are both not only a copyright holder but we also a user of the copyrighted works as well. So we need to keep the balance you know between the two natures and uh because we are expected to provide you know uh information and serve the public. So in order to but archive is probably one of you know uh carried out for the purposes as well. So uh we need to keep the balance when you think about you know how whether you would like to how can I say protect your archite you know works uh and you need to have a kind of a internal you know discussion thoroughly uh from various kinds of aspects including the nature of the broadcasters.
I also would like to quickly add on about the current situation uh on this question.
Yeah, I also want to quickly add on this. Uh this is one of the key question actually on the input side training large language model and actually currently the most uh if you track the AI gen AI and the copyright case in the court actually this is one of the most uh dispute area and uh currently uh to my knowledge US got the most the cases are lot increasing you know and start from um we we need to also uh considering like uh um some some like comedian even like sue the big tech company and like New York Times some big company together to shoot to protect their copyrighted works and I would like to say that um currently uh the there's still pending in the court in the US it's pending in the court the judge need to decide whether is belong to fair use whether this belong to copyright exception you know and also um eventually I think a lawmakaker have to be more clear about this training large training large language model use copyright works how it is this uh fair use is is this uh need to compensate the copyright users corporate holders actually I saw a lot of you know debate and attention and uh and I I also see that different country jurisdiction maybe they have different concerns um some some country may say hey we want in some way to make sure the tech company have enough data to train the uh models and some maybe worry about the human creativity and then maybe more turned on protect the copyrighted works and personally I think that um I I think that um for the company if they want to protect their copyrighted works they can use several ways such as um in order to wait the law is more clear uh you limit the assess. So all your archive if you really don't want to people use it maybe you can control the assess and also they're talking about opt out uh eventually I think uh in the future we'll be more clear on this and also maybe make a deal and bargaining about if you want to use this to training the the the model and how to what is the the remunation uh unfortunately some of the foundation model already trained so so and how and also in practice how to prove there is a infringement is another difficult issue actually if you see the case okay that's uh what I can comment on yeah >> thank you Sarah I think uh it is about time that we wrap up our first session thank you to all our speakers and if you do have more questions um you can actually bug them uh during the coffee break I'm sure and uh you could always uh email Sarah regarding uh questions that you want to ask her.
Um I will hand over the proceedings to Dr. Siman.
>> Let's just give a round load of uh applause to our speakers. It was a very very insightful session. Thank you so much. We will now all break for coffee and if we can reconvene at 11:40. So just 10 minutes within 10 minutes we need to be uh back because or else our later sessions will not fall in place.
Thank you so much.
>> Thank you.
>> Thank you. Thank you so much, Miss Hojan. Thank you.
Mr. Marshall Sihan, Mr. Indra Singh, Miss Yuan and Mr. Awan Aziz and Miss Galina to please come on the stage.
Speakers for the second session, if you could please come to the stage.
Thank you so much.
All right.
Okay. Okay.
Good afternoon everybody.
Welcome to the second session of uh the AU IPLC conference.
So my name is Caris. I'm from Zara Media Group and I'll be the moderator for the session uh right now. So the first speaker is Mr. uh Marcelan the chairman of the National Collective Management Organization for Related Rights Honors. So uh Mr. Marcel, please proceed.
>> Okay. Good afternoon, distinguished chair, colleagues and friends. Uh first of all allow me to thank ABU intellectual property and legal committee and as host for inviting Alam Cayenne into this uh very timely conversation on AI copyright royalties in the digital content economy. Again it is a privilege for us to share Indonesia's perspective on a topic that it at once very technical and also very deeply human. We need to ensure that in an era of streaming simultaneous cast, webcast, OTT and now AI the value created by music still reaches the people who create it. So my assigned sub theme is to ensure fair remuneration the role of LMAN and music royalty schemes in the digital and AI era. So I will approach this from three different angles. First is how artificial intelligence and platforms are changing the royalty landscape. The second how LMAN Laga Management Collective National is rebuilding royalty governance so that it is fair, transparent and adaptive.
And the third how these efforts are being anchored in Indonesia's copyright reform and in regional cooperation.
Okay.
Next.
Okay. Let me begin with the changing landscape. So around the world, digital transformation has changed how music is produced, distributed, used, consumed, and increasingly even created.
So music is now routed through streaming platforms, video sharing sites, social media, live streaming, digital radio and digital karaoke. And at the same time, artificial intelligence systems are trained on fast data sets that may include music and they are increasingly used to generate, process, and recommend content.
The result is a tension that many of us in this room also recognize. On the one hand, the overall economic value of music is growing. And on the other hand, the distribution of that value is not always enjoyed proportionally by creators, performers, and right holders.
In this environment, the traditional enforcement model of copyright, which is like waiting for an unauthorized copy and then suing for infringement, becomes increasingly insufficient.
So the practical question is no longer only who is the owner or is there an infringement.
It becomes can we see what is being used? Can we license it? Can we enforce obligation and can we distribute the money back to creators and right holders in a way that they understand and trust?
Okay.
In practical terms, the artificial intelligence challenge is no longer abstract. It includes training data sets that make content that may contain music, AI generated songs, voice cloning that imitates performers and style imitation that reproduce the recognizable creative identity of composers.
Each of these raises not only question of authorship of infringement but also questions of licensing, attribution and remuneration.
So when we discuss AI and music, we should be careful not to treat it only as a technological issue but also a governance issue, the market issue and remuneration issue.
Against that backdrop, I would argue that legal certainty in the digital NI era rests on four pillars. The first one is clarity of rights. who owns what and in which layer of the work. Second, transparency of use. Who uses what, where, and under which terms.
Third, fairness of remuneration.
Who pays how much and how that value is shared. And fourth, enforcability of obligations. What happens when someone chooses not to comply?
Without all four pillars, protection remains declaratory on paper. This is important. Remains declaratory on paper but weak in practice.
This is why so many issues now converge on a common practical questions. Who bears responsibility and who gets paid in a digital content economy shaped by AI and platforms?
Let me now briefly situate Indonesia.
Under the law of 28 of 2014, Indonesia has a comprehensive copyright framework.
Protection arises automatically once a work is expressed in tangible form.
The law recognizes both moral rights and economic rights and acknowledge multiple layers of rights in music including the right of authors and composers, performers, producers of sound recordings and broadcasters.
This give us a solid normative foundation. However, that foundation was designed for a world before large scale of AI training and before the current dominance of streaming, simalcast, webcast, OTT and other digital platforms.
This is where LMIN comes in. So, LMI must be understood not only as a collector and distributor of royalties, but also as a collective licensing institution, a provider of usage data and a bridge between music users and right holders.
And even as technology changes, the principle does not the use of work still requests a legal basis. Economic rights remain attached and creators remain entitled to economic benefit from the exploitation of their works.
In this invitation we received Alam Caen was asked to speak about ensuring fair remuneration. So for us that phrase carries three attached adjectives.
It must be fair.
It must be transparent and it must be adaptive.
Fairness concerns how value is allocated among authors, performers, producers and other right holders.
It requires clear rules, consistent, consistent application and alignment with the real economic pathways through which music generates values.
Transparency means that creators should be able to see the usage data.
They should be understand the distribution rules and trust the management of the money collected.
Adaptive means that the royalty system cannot remain trapped in the analog era while a value moves into digital and AI shaped environments.
This adaptive dimension is particularly important because music now generates value across a much wider field than traditional broadcasting alone. So when we speak of digital uses, we should think not only of streaming in the narrow sense, but also of video sharing platforms, social media, live streaming, digital performance, digital radio and digital karaoke. These are all environments where music creates economic value and where royalty logic must increasingly follow actual usage.
For Indonesia, this mean that LMAN's licensing horizon must explicitly include streaming, simultaneous cast or simalcast, webcast, OTT services, online radio, social platforms, and other hybrid digital uses.
If these new channels generate value but remain outside the effective royalty system, fairness will collapse.
At the international level, several issues are already being discussed more explicitly.
So these include data set licensing, AI transparency, disclosure obligation for training data sets, collective licensing and even extended collective licensing.
Indonesia does not need to copy any model mechanically but it does need to prepare an adaptive governance framework that can engage with these developments.
This is important because AI models, platforms and content markets are all crossborder by nature.
National system must remain rooted in domestic law but they cannot be designed in isolation.
This bring me to the ongoing revision of Indonesia's copyright law. Public information about the draft indicates that it will address AI and royalties more explicitly. In broad terms, three elements are particularly important.
The first is AI related provisions. The draft direction reportedly acknowledges AI assisted and AI generated works and begins to frame training uses and outputs.
The second is platform duties. Digital platforms are expected to pay royalties, share usage data, cooperate with collective management bodies such as LMKN and support near realtime reporting.
The third is enforcement certainty.
The law seeks to clarify who clarifies as a user and who is the obligated payer across different business models.
Again for Alam Cayenne this definitional clarity is essential.
We cannot enforce what we cannot divine.
But diagnosis alone is not enough. We also need a road map.
From our perspective that road map has four dimension. First regulations, technology, institutions and governance.
Regulations includes AI, legal certainty, licensing, design and public education.
technology includes data set transparency, national metadata, digital fingerprinting and an interoperability of data system.
Institutions include strengthening Alen and collaboration among collective management bodies.
Governance includes structured cooperation with digital platforms and the development of AI ready royalty administration.
This effort must also be sequenced. So in short term regulatory revinement and public education are necessary.
In the medium-term we need integrated national data and metadata standardization.
In longer term the goal should be an AI ready royalty system and a fully functioning digital licensing ecosystem.
Let me close by returning to the regional perspective of ABU and IPLC.
So, Indonesia's experience can contribute at least three things to this dialogue. First, a concrete example of a national collective being re-engineered for a datadriven digital environment.
first. Second, a life case study of how AI issues, platform obligations and copyright reform can be integrated.
The third, a commitment to regional cooperation, recognizing that no country can solve these issues alone.
When we speak of ensuring fair remuneration in the digital and AI era, we are not only talking about money.
We're also talking about trust.
visibility, accountability, and workable compliance.
Technology will continue to evolve, but respect for human creativity must remain the foundation of a sustainable music ecosystem.
So for Indonesia, for us the path forward is to move from declaratory protection to operational protection so that rights are not only recognized in law but can actually be licensed, monitored, paid, verified and enforced.
Thank you very much.
>> Thank you Mr. Marcel. Our next speaker is Mr. Indra Singh, director of ABU news.
You're done.
>> Thank you very much. Well, I feel a little bit of an alien in the new in the in the room because I don't come from a legal background, but we deal with legalities on a daily basis. And you know when when we talk news there's no way that news can go forward without having the legal framework in place. So before I get going and talk a little bit about Asia Vision and what we do and how we counter this, it'd be good to play that video that I have got. It's not a very long video. It's about 4 minutes long. if they can bring that up.
>> Thank you.
>> Of truth or realism.
>> There may be no returning the AI genie to the bottle.
>> If we start to push out AI generated content that changes the narrative of a story, we are no better than any bad actors out there.
Greetings. I'm Alex Golden. I'm the senior director for the product team at NBC News Group and welcome to NBCU Academy.
In terms of defining artificial intelligence, when used properly, it is the automation of human intelligence. It should make life easier for things that currently take a lot of manual activity to do.
We've been involved with artificial intelligence and machine learning for a good number of years. We started off with something as simple as transcriptions. People in there transcribing in a room off of video. We thought that we could also cut costs by getting an AI solution that was more self-s served. They put it into a system as an MP3. Get it back. It's fully transcribed. The other side of that's capturing the data of pictures, the picture to text, Biden at the podium, Trump with a red hat. Those are things that currently would be manually logged.
The thing that always comes up is whether AI is a quote unquote replacement technology. We don't see it that way. We see it as an assistance technology. Our charter for using AI as a company is to have it help and improve workflows and where it fits the use case replace workflows but not replace people. It should always have that human element of being monitored, being attached to some amount of human interaction or workflow so that we make sure it's being used the right way.
The big elephant in the room becomes generative AI. And obviously that's the huge buzz on the block.
>> So how can I help?
>> Generative AI is AI that can create text, images, and media from human commands. Go ask it to do something and it'll do it. You can talk to it.
>> From fake pictures of Donald Trump getting arrested to a fake video of President Biden instituting a military draft.
>> If we use something on air that was faked, that's a huge potential risk, of course. But the other risk is if somebody takes our content out there in the public space and changes it. What if somebody were to take Jim Kramer's words, >> buy change them around?
>> Sell, sell, sell.
>> That could affect markets. There's actual real world ramifications.
>> Moving forward, we need to be more vigilant with what we trust from the internet.
>> A lot of our editorial teams are using very timeconsuming activities to be able to check for fake content. What AI can do is quicken the process. We want to use fire to fight fire. Employ AI as the best way to sus out the AI being used by bad actors. There probably is more upside on how journalists can actually use it. Whether it's to take large quantities of text and summarize to generate conclusions based on meetings, upscale images. These are all things that I think journalists are starting to tackle now as a way to have this AI improve the work that they do.
>> Don't use AI in a vacuum. Make sure that you have brought up this attempt to use AI to your own team leadership. Don't let AI do the work for you in such a way that you then become the self-perpetuating replacement of your own job. This is serious breaking news, but a little behind the scenes, a little a little flavor. I'm with one of the most legitimate correspondents on Earth.
>> Have fun and experiment with what is out there. It could be fun to play with. For sure.
>> AI is not only being used to clone actors faces and voices. It's also being deployed to do CGI and visual effects faster and easier than ever before.
>> Should journalists be afraid of AI?
No. But should they be cautious about it? Absolutely. We still believe that journalism is done by journalists. But if we're going to use AI, we still need to maintain the editorial best standards and best practices. Those don't go away.
AI cannot be used as a shortcut. It's the exciting new stuff that we're working on that excites me the most. I'm Alex Golden with NBCU Academy. Thanks for watching.
So that's just a uh little clip from one of the uh traditional broadcasters out of the US. So at Asia Vision, as most of you know, we deal with hundreds and hundreds of news uh content particularly from our broadcast partners on a daily basis and with our sister unions as well, which ranges from Europe to Africa, Caribbean, uh to Latin America and the Arab states.
Does AVN have a role or responsibility of what is loaded by our members is the question we keep on mulling about and talking about and many of times I know Sema and previously others have worked on it. If the content is being loaded by our members and then being used by another member, who holds the responsibility?
Who is at risk? Should it be deep fake or should it be content that is not um true?
It's it's a fine line because AVN is the facilitator through ABU.
However, it is our responsibility to ensure that whatever is pumped out through the platform is checked and thoroughly checked. And that is where we try to work with tools to be able to identify. Mind you, I'm not saying that our members do pump out fake news or fake stories, but there are times and things that come into play is using images, perhaps AI generated images in news content, but not declaring it. Um the a good example is the recent in February when we had the um Iran war breakout. Um there was so much content that was being put into the system not only from our members but from our partners as well.
Many of those content that came in, particularly from one or two of our members, was being questioned a lot by fellow members because it had content that was not readily available elsewhere. And it sort of was being looked upon as propaganda material or something that was created using AI tools or created using uh different different tools that are available. Now, how did we manage to counter that? That is where AVN's role was very important.
Anything that did not fit the bill of what we try and promote was totally taken out of the system. We could have gone through this whole process of probably trying to identify each and every single frame of the footage, but as mentioned by NBC, the the video we watched, that means more human time to sit down there and look at hundreds of hours of footage. And at Asia Vision, we don't have that type of manpower.
That I'm just giving you an example of what happened at the Iran war. Then couple of our members would load something on the platform without restrictions or without source. And if someone like the EBU picked up upon it, their first question would be where was this sourced from and if it is genuine.
So that meant that we had to before it got published um look through all of those to be a trusted news exchange. I think it's very important and I come from a perspective where I've worked I had a I've worked with uh Rossa at my previous workplace and before that we never had a legal team in place at the Fijian Broadcasting Corporation and we had a number of cases whereby even though AI was not you know at that r stage already but there were other issues around picture usage as around image use around reproduction of images used from elsewhere.
I I I I feel if news rooms do not have legal expertise in this day and age more than ever, we are in for a for a rough ride. I think it was in the first session, if I'm not mistaken, the talks about cases ending up in the US courts, number of cases ending up in courts around the world. Newsrooms very soon will have to start answering because the third party users do not care about things and that is where our responsibility as news directors as heads of news organizations is very very important. With that also comes how much of a legal training are we giving the news people in its individual newsroom or organizations or unions. That is where and I I come from an experience of having worked in a newsroom, having led a newsroom where many of times legal is seen as something like you know totally not part of our team because legal might tell us don't do this story because there's a loophole there which could come back and you know and and land us in trouble. But the mindset needs to change. And later this year at the global news forum and earlier this year at the coordinator's meeting, we were fortunate to have Dr. Sema. later this year to be hosted by the ABC in Sydney. We have a session there of news and legal aspects of how we do things because for me it is very important that each and every member or each and every newsroom at least within our Asia-Pacific Broadcasting Union understands that just like Asia Vision just like the individuals we need to be working together as a team.
Now the other thing is it's your reputation. So trust and truth is very important because if we as Asia vision or Euro vision or ASBU etc the sister unions if we are not able to provide the trust and truth to our members by getting involved in this uh AI era we are losing the trust not only of our members but of people as it is uh people are switching off traditional media.
they are going across to digital and all other sorts of media. So it remains our responsibility of ensuring that we have that from from content responsibility point of view. It is uh very important to have a set of rules in place for each of the newsrooms and I know most of the newsrooms do most of the broadcasters do but you'll be surprised that some of our members don't have an AI policy in place for newsrooms and they're very honest they have spoken about finding it tough to legally have a legal or not really legal but to have the rules in place or have a um have um how trying to find a word here to to ensure that the journalists and the newsroom is able to use that platform to refer to should they be in any dilemma to be using content. I think that's where we can work together. We've got expertise here and our members range from very big organizations to some very smaller ones around the Asia Pacific and beyond. I think if we can work together to help each other, we will be fighting this.
It's just a small step. I totally agree with I think it was earlier on trying to get an MOU or trying to just work with each other. That is going to be very much. So just finally before I wrap up, I think one thing I'd like to say is newsrooms and news people need to change their mindset, involve legal, work with legal to ensure that when that call comes one day or writer one day, you are in a safe place because you've done your 101% check rather than just going because you wanted to break news and ended up not having the, you know, original content, etc. Thank you very much.
>> Thank you Mr. Indra.
Our next speaker is Miss Yuette uh the corporate officer from China Media Group.
Good morning everyone. I am Yuen from China Media Group. uh in China generative AI is rising quickly.
Everybody almost everybody use AI to assist the life and work and study. Just as Desmond mentioned students they don't ask teachers questions they just ask AI and AI give them answers.
Um but nowadays um especially recently because of the deep fakes and ethical risks brought by AI in China actually people are more cautious now. Um for example when when people use AI in program production or or or other work we are now in use it in a more careful and more cautious way um comp we are actually people are coming down compared to the previous excitement pure excitement. So um here I have a I have a little story. Uh there's a place uh uh there's a place there are many big mountains and uh there grow many kinds of mushrooms and local people love mushrooms very much. But the question is that some mushrooms are poisonous, some mushrooms are not poisonous. And a group of people they went to hiking, go hiking in the mountains and they asked AI they they found a bundle of mushrooms and they asked AI are these mushrooms poisonous or not poisonous and the AI answered not poisonous and they collected the mushrooms go back cook them and then turned out poisonous they were all sent to hospital. So this is a true story and that's why people are now getting more cautious now. Uh today actually I'm going to provide a a case study of uh sports copyright issue. I was planning to talk about the world cup actually because the world cup has just concluded and it was very hot topic and China media group has made a very successful copyright negotiation with FIFA one of the best negotiations with FIFA in the in the past history but we are still doing the summary work yet I don't get the data yet so um uh I video choose to talk about last year's the world games 2025 Chungdu actually our copyright work are more or less the same. So my um the this this slides the slides consist of three parts. Part one uh brief to the sports event. Part two our copyright protection for the games and achievements. Part three joint efforts by various sectors of Chinese society to build a variable environment for copyright protection of sports events programs. Now part one brief introduction from August 7th to August 17th the 12th word games was held in Chungdu Sutran province of China. This is the first time that the games has been hosted in Chinese mainland. So it has great significance for the Chinese people and at the highest level international comprehensive sports uh uh for nonolympic sports the world games and the Olympic games actually they complement and reinforce each other nearly 4,000 athletes from more than 100 countries and regions around the world gathered in Chungdu.
This is a the largest one with the most athletes participating in its history.
Among the athletes, the oldest is a 68-year-old archer and youngest is then only 13 year old squash player. Uh they competed in 34 sports disciplines and over 250 events with a total of 256 gold medals awarded.
Uh next next slide.
Uh yes, just here I have a video to show you. Please pay.
Yeah.
for Thank you for watching. That's our CCTV 5 uh sports channel uh broadcasting of the games. Now let's come to the the copyright business. First is acquisition of rights. Some Chinese experts have been doing research in this field and they hold that there is a sports event media copyright industry chain the with the upstream owners and operators of sports events including IOC, FIFA, NBA and so on the midstream sports media uh that means broadcasters internet social media and the downstream consumer s derivative product cons uh producers. And before broadcasting a sports event, we usually sign a contract with the event organizer or its agent to get the rights we need, including rights for broadcasting, sublicicensing, and derivative goods. In accordance with our contract with um the the executive committee of the world games, we acquire the ex exclusive all media rights and exclusive sublicicensing rights for the games in Chinese mainland, Hong Kong, Macau and in it's entitled to take antipariracy actions against any infringement.
Uh now please uh have a look good uh quick look have a quick look at uh introduction to China media group.
Yes. Next please uh just quickly go through it. Thank you.
Given the big international significance, social value and economic value of the games, we attaches great importance to the copyright work of this event. Um before the opening ceremony.
Next slide.
Next slide. Uh next. Yeah, here. Thank you. Before the opening ceremony of the games, we issued multiple copyright statements to clearly declare the copyright we own were enough infringement risks and delivered early warning letters to potentially infringing platforms for public notice.
This have laid a solid foundation for our later work. And during the games, we deployed dedicated working staff to closely follow the event schedule and implemented uh seven multiple 24 realtime monitoring to promptly detect various infringement acts. In terms of ondemand service, short video infringement was the most prominent.
Most of such infringing content was uploaded by users featuring no costs, fast dissemination speed and the ability to gain traffic in a short period. Meanwhile, there existed quite concealed at diverse infringement methods. So nonauthorized platforms provided live streaming and on demand services of the program uh through deep links. In response to this, our team worked on duty 24 hours every day, taking strong measures throughout the whole process with attention paid to all aspects. We conducted comprehensive monitoring and real-time handling of infringing content across the entire network, collected full evidence against the platforms, especially those social media platforms with in severe infringing acts. And we delivered lawyers letters immediately to defend our rights. And after the clothing ceremony of the games, after the clothing ceremony, we pursued legal responsibilities of infringing entities in accordance with the law. For entities that committed serious infringements with severe impacts, we filed administrative complaints or initiated legal proceedings to um to to fulfill uh to pursue their legal responsibilities to the fullest of the law.
Joint efforts have been made and practice achievements are seen in China.
Um yes please uh quickly quickly next one and next and next yeah here thank you uh professor Wong from the law school of uni university of China he made an in-depth analysis of the practical difficulties in the copyright protection of sports events programs he current sports events program shows three major copyright characteristics. The first one is high co high copyright acquisition asks uh costs high copyright acquisition costs over the past decade the costs incurred by Chinese domestic entities in purchasing copyrights for foreign sports events have grown obviously that's a very high burden and second complex history history of legal disputes before the third amendment to China's copyright law in 2020. There were repeated disputes over issues such as whether sports events programs are copyrightable and the legal definition of online live streaming. These were basically resolved after the amendment to the law. And number three is significant differences in copyright terminology. There are actually barriers to the conversion of legal terms between China and foreign countries.
because we have different language systems and different academic systems.
So as well as between the uh as well as between the legal commu community and business departments in China which may easily lead to disputes. Professor Wong cord on the academic and industrial circles to strengthen cooperation promote the unification of legal interpretation and industry practice to adapt to new challenges in digital age.
Uh, as CMG had long been engaging in the copyright protection of major sports events such as the Olympic Games and the World Cup, we have uh invested a large amount of human resources and money and other resources in the broadcasting. So we have a deep understanding that copyright protection is directly related to sub uh sustainable development of the industry.
The protection of sports events program in China has benefited from the joint support of all sectors of the society.
However, in the face of diverse patterns of violation and the delay of platforms in fulfilling their responsibilities, it is still necessary for us to work together to stimulate the innovative vitality of the industry. Especially now we are in the new AI era. Uh new challenges are facing us in the future.
So we there are still a long way to go and that's all for my sharing. Thank you for your attention.
>> Thank you Miss Yuen.
Our next speaker is Mr. Awan Azis the finance manager of AVU.
testers. Salam alaikum. Uh good afternoon everyone. Distinguished chair, respected speakers, ladies and gentlemen. Uh first of all, I would like to thank ABU legal department, Miss Dr. Sima and uh our colleagues at RRI for inviting me to participate in this conference.
Um just like Indra mentioned earlier uh looking at this uh today's program I I have to admit something I felt a little bit uh probably the odd one out because most of the speakers here are legal practitioners I could say IP specialists maybe copyright experts I am not. I come from finance.
So at first I wondered what a finance manager could possibly contribute to a room full of legal experts.
Then I realized something perhaps uh this is precisely why I was invited because once copyright is created, protected and licensed, someone needs to ask a question.
How do we turn those rights into sustainable value? Ladies and gentlemen, that that is where I think finance comes in.
Throughout this morning's uh session, we have heard about copyright protection, legal frameworks, emerging technologies such as AI or as someone mentioned this morning, generative AI.
All of these discussions are important.
No doubt about it. Without copyright, there is nothing to protect.
Without legal certainty, there is no confidence to license or commercialize content.
But I would like to offer another perspective.
A copyright or a contract or a license does not automatically create value.
Value is only created when organizations are able to govern those rights effectively.
Allow me to share a simple observation.
Many organization spend enormous effort negotiating agreements. Sometimes months, weeks, massive efforts. Everyone celebrates once the agreement agreement is signed.
But after that, who monitors whether the license is being properly used?
Who checks whether the royalty payments are received on time?
Who measures whether the agreement actually delivered the commercial benefits that were originally expected?
And who asked whether the investment made any financial sense?
Those questions rarely receive the same level of attention as the contract itself.
As someone from finance, I see governance as something much broader than compliance.
Many people think governance means more rule.
We saw with Marcel's presentation earlier on governance.
Some some might think governance means more approvals, more procedures, layers and layers and layers of compliance.
But I see governance a little bit differently.
To me, governance simply means ensuring the right people making the right decisions using the right information at the right time.
In fact, let me share you a recent experience.
Only recently I sat in a meeting discussing the acquisition of a broadcasting rights for a major sporting event.
Around the table were colleagues from legal operations, finance of course and the management. We were all looking at the same project but each of us was asking different questions.
Legal was considering the contractual obligations.
operations was focused on delivering the event successfully.
finance of course we were asking all those hard questions like uh what's the total investment what are the risk involved what are the ROIs blah blah blah so that meeting reminded me that good governance isn't about one particular unit or department having the right answer it's about bringing different perspectives together before a decision is being made that is why I believe governance is ultimately about collaboration ation instead of compliance.
To me, governance is like a steering wheel, not a speed bump.
Having said that, this is why I believe IP should not be viewed merely as a legal asset.
It is also a commercial asset.
It is a strategic asset and ultimately is a financial asset.
In an era where traditional broadcasting revenues are under pressure, intellectual property may well become one of the most important assets organizations possesses.
Protecting IP is only the beginning. The greater challenge is ensuring that those rights continue to generate value for the organization and its stakeholders over many many years. That requires collaboration because legal cannot do it alone.
Finance cannot do it alone. Nor do operations nor do technology.
Success eventually comes when these functions work together. Each contributing to its own expertise towards a common objective.
There is one more lesson that I come to appreciate.
Good governance is not only about making the right decision.
It is also about ensuring that important decisions are made early enough so that each of these functions to contribute meaningfully because when difficult discussion only begin at the final stage of a project.
Organization al often find themselves making important decision under unnecessary time pressure.
Good governance, ladies and gentlemen, gives every function the opportunity to contribute before options become limited. That in my view is how we make better decisions. Before we con before I conclude, allow me to leave you with three simple thoughts. Not as a finance professional, not as an accountant, but as someone who believes that good governance is everyone's responsibility.
The first is this.
Every IP, every intellectual property asset should have a clear owner. Not necessarily the legal owner, but someone within the organization who is accountable for ensuring that the asset continues to generate value.
Too often IP becomes everyone's responsibility which eventually means it becomes nobody's responsibility.
Someone must ask are we still using these rights? Are we still protecting them? Are we still generating value from them?
Because if nobody asks those questions, valuable IP property can quietly lose its commercial value over time.
Second thought is this.
Every major licensing or rights agreement should have a business case.
Notice I did not say a legal opinion nor did I say a financial model. A business case.
A good business case brings together different perspective. Legal asks can we do this? Operation asks can we deliver this? Finance will ask can we afford this? And ultimately management will then ask the most important question.
Should we do this?
Those are very different questions.
But I believe from my perspective, good governance happens when all of these four questions are answered before the decision is being made.
My final thought concerns AI. We talk about it a lot this morning and I have to be frankly honest. I am not an expert in AI but AI is changing almost every aspect of our industry not just broadcasting industry but probably in our life.
It generate contents analyze contract can identify copy copyright infringements. It can even even assist organization in monitoring the use of licensed content. These are obviously exciting developments, but AI does not remove the need for governance. In fact, I believe AI makes governance even more important.
Technology obviously can provide answers, but it is only us can decide which questions are worth asking.
Technology obviously can process information but only us can balance commercial objectives, legal obligation and organizational values.
The future obviously will not belong to organization that simply adopt AI but it will belong to organization that combine AI with good and robust governance. As an analogy, I look as AI as a big engine of a car, but governance is the steering wheel that decides where the car wants to go. Allow me to finish with one final reflection.
Earlier today, we have heard many excellent presentation explaining how intellectual property should be protected.
I fully agree, but perhaps we should ask another question.
What happens next?
Because once we have successfully protected our intellectual property, we have to ask ourself does it support innovation? Does it strengthen collaboration? Does it create sustainable value for our organization?
Because if the answers to this question is a no, then perhaps that's where governance truly begins. Ladies and gentlemen, as finance professional, lawyers, broadcasters, managers, we all see IP from different perspectives, different lenss, that diversity should not divide us. It should strengthen our decision- making.
Good governance again is not about one unit, one department being right. It's about everyone contributing its expertise so that management can make better decisions. Perhap that is the real value of conferences such as this.
Not simply exchanging legal knowledge, but learning how different professions can work together to protect, manage, and maximize the value of intellectual property. Thank you for your attention.
>> Thank you, Mr. Aswan. Our next speaker and final speaker, Miss Galina Doescu, head of copyright and related rights at Radio Romania.
Okay, now it's okay. It's okay.
Okay, good afternoon everybody. Uh, first of all, I would like to express my gratitude to the host for inviting me to the 30 second uh, Abu Intellectual Property and Legal Committee meeting. It is a great privilege and an honor to be here and uh today and to share with you our company's experience regarding one of the most challenging aspects of the European artificial intelligent act. The transparency use of deep fake technologies for for public service media. The most relevant provision is article 50 paragraph 4.
This provision establishes transparency obligations for deployers of AI systems. In other words, it applies to organization that use artificial intelligent to create or manipulate content, not only to the organization that uh develop AI systems.
Public service broadcasters clearly fall into this category whenever they use AI tools to uh in their editorial activities.
Before discussing uh our institutional experience, let us briefly clarify what the term deep fake actually means.
According to the European legislations, one of the strength of the artificial intelligent act is that it provides a legal definition of a deep fake.
According to AI act, uh, deep fake means AI generated or manipulated image, audio or video content that resembles existing persons, objects, places or other entities or events and would falsely appear to a person to be authentic or truthful.
This definition is particularly uh important because it contains three key elements.
The content must be generated or manipulated by AI. The previous slide, please. Uh the the content. Yeah, the previous uh maybe I could u try.
>> Yeah. Uh let me uh uh Okay.
Uh it doesn't work.
I tried to no.
Okay.
Yeah, this one. So, this definition is particularly important because it contains three key elements.
The content must be uh generated or manipulated by AI. There must be significant resembles to reality. There must be a risk of misleading the public.
Deep fake can include image, audio, video or multimodal combinations.
According to article 50, disclosure must be clear, visible, easily accessible, easily accessible provided at the time of exposure. European legislators provide one of the most important clarification.
If the cont if the deep fake is used in an artistic, creative, satirical or fictional work, uh the disclosure uh the obligation of disclosure remains but in a proportional manner. The purpose is not to harm the creative value or the creative experience.
uh for instance a feature film that recreate a dis uh diseased actor that this must be disclosed but in a proportional manner.
Now I would like to uh talk about a section that is very extremely important uh for the for the press and public communication.
AI generated text on matters of public interest.
Article 50 paragraph 4 also applies to AI generated text that are published to inform the public on topics of public interest such as politics, public health, economy, security, elections, crisis, public policies.
When does this obligation for text apply? Not every AI generated text falls under this obligation. The following uh criteria must be met cumulatively.
The text must be generated or substantially modified by AI. It must be published.
The purpose of publication must be to inform the public. The topic must be of public interest. For instance, an article about election written by AI and public uh and published online falls under the obligation.
Uh an economic report generated by AI for the public falls under the obligation. An internally uh company email uh does not fall under the obligation. There is an important exception if the text is subject to editorial control and editorial responsibility.
Artificial intelligence is capable of producing highly realistic voices. It can recreate uh public figures. It can generate synthetic interviews. It can create images that look entirely authentic.
This is precisely why article 50 introduces transparency obligation. The purpose is not to bend the innovation or to discourage uh public service broadcaster uh for using AI. Rather it it is to ensure that citizens know when artificial intelligence has played a significant role in creating or manipulating content. In this way, transparency serves as a safeguard uh for informed decision making. This brings me to our own institutional experience.
Although the legal obligation uh set by article 50 paragraph 4 are only now becoming operational within the European regulatory framework, our organization decided not to wait until legal compliance becomes mandatory.
Instead, we choose a proactive approach.
Currently, artificial intelligence is already being used within our organization to generate illustrative images for certain editorial or communicational purposes. These images are not intended to replace journalism.
They do not replace documentary photography. They do not replace editorial verification. Their role is purely illustrative. We believe that transparency should accompany their use before the deadline in the AI act become applicable. For this reason, uh every AA generated image published by our organization is clearly labeled as being generated using artificial intelligence.
Thus, starting from mid July, all images created with AI are labeled and those created before mid July of this year are scheduled to be labeled as well. As you can see, um the two images marked with AI label were used by our employees to illustrate the article written and posted on our station's websites. Of course, under each photo, it states that they are for illustrative purposes because we believe that transparency should not start with regulation.
Transparency should start with institutional culture. One of the uh first practical step taken by our organization was drafting an internal code on the responsible use of artificial intelligence. The purpose of this document is not simply legal compliance. Its purpose is to establish common organizational principles. It provides guidance on the responsible use of artificial intelligence, editorial transparency, human oversight, accountability, and safeguard public trust. Our goal is to ensure that artificial intelligence is used consistently across the entire organization rather than deferring from one department to another.
However, documents alone don't create an organizational culture. People do.
Artificial intelligence is continuously uh continuously evolving. Therefore, uh I forgot to tell that this code also uh contains an annex uh with uh models of the AI uh uh this code also uh includes an annex with the AI labels applied by our employees were appropriate uh that to inform the public when AI is being used. However, documents alone as I told before don't create uh create an organizational culture. People do uh and for this reason our organization uh artificial uh artificial intelligence is continuously evolving. Therefore, journalists need continuous learning, editors need continuous learning and manager need continuous learning. For this reason, our organization is preparing a structured training program dedicated to artificial intelligence.
The goal is not to turn journalist and AI uh specialists. The goal is to help every employee understand three simple question. Where can AI be used? When should transparency be applied? And who remains responsible for published content? These questions seem simple. In reality, they define the future culture of AI assisted journalists.
Ladies and gentlemen, let me conclude let me conclude with a simple reflection. Throughout it its history, public service broadcasting embraced many new technologies. radio, television, digital uh broadcasting, online platforms, social media.
Artificial intelligence is simply the next chapter in this evolution.
Therefore, the real challenge is not technological, it is institutional. How do we preserve public trust while embracing innovation?
Um, I believe that article 50 provides an important answer, not because it introduces uh new administrative obligations, but because it reminds us of something journalist has always known. Transparency create credibility.
Credibility create trust. And trust is the foundation of public service media.
Artificial intelligence will continue to evolve. The technology will uh undoubtedly become more powerful. But one principle should remain unchanged.
Editorial responsibility belongs to human and public trust remains the most valuable asset that any public broadcaster possessed. For this reason, I would like to close with a sentence that in my opinion sums up the entire philosophy of artificial intelligence. Artificial intelligence can help us create content. Transparency helps citizens trust that content. Thank you very much.
>> Thank you, Miss Scalina. uh such wonderful presentations from our speakers in this session. I'm sure they have a lot more to share with us. But unfortunately, we do not have time for Q&A. So if you have any questions, you may approach them during lunchtime. So uh thank you everyone. Uh we will now adjourn for lunch. Thank you.
>> We can give a round of applause to our uh speakers. That was a very very insightful session. Thank you so much.
We will now break for lunch and we will reconvene at 1:35.
1:35. So we have around 35 minutes of lunch. All right. Thank you so much.
>> The lunch is being served outside.
Heat. Heat.
Down.
Down.
Heat. Heat.
Heat. Heat.
Down.
Down.
Hey, hey, hey.
Heat. Heat.
Hey, hey, hey.
Baby.
I don't know.
If I had to live my life without you, the days would all be empty.
This to the stage.
Okay, shall we?
Okay, good afternoon everyone. Uh, I hope all of you had a good lunch. Uh, more importantly, how was the Was that nasi padang? Just know we had was it nasi padang?
Huh?
How was the nasi padang? Anyway, anyway, before we continue, I'd like to extend our sincere thanks to Andy and the RARI team for having us over lunch.
Now, let us move on to session three.
Um, for those whom I haven't met, my name is Azwan. I am from the ABU Secretary Office in Koala Lumpur. So, it is my pleasure to moderate this afternoon session.
Uh let us first uh introduce the speakers.
On the left of me is Miss Jacqueline Ford from ABC and then Mr. Alex Raza from RTM and uh the third will be Ashwini Natison.
So uh just a small reminder we have about approximately 15 minutes per speaker and um I'll give you a heads up um when we reach the one minute times.
Without further ado uh ladies and gentlemen uh please join me in welcoming our first speaker for today's session.
Jacqueline, >> hello.
>> I'll just pop this one down.
>> Hello everyone and uh thank you very much for the invitation to speak today.
I'm presenting from the perspective of a lawyer and governance professional working at the Australian Broadcasting Corporation.
What I want to offer today is not a technology briefing or a legal briefing from or a briefing from a journalist involved in content and editorial output. What I want to offer is a governance and accountability lens to the adoption of AI, which I think is critically important to maintaining trust in public media and the content that it produces.
We're all operating in an environment where AI tools are probably already embedded in our organizations in many cases without a considered governance framework that is fit for purpose.
The question I want to look at is what does fitforpurpose actually look like for public media? How can we safely use AI to support trusted news, cultural content, public interest journalism, and universal access? I'll be drawing on Australian perspectives, which is what I know, and I'll let you know when I'm being specific about the ABC.
Before we talk about governance, it's worth being precise about what we're governing and the risk if we don't get AI governance right. This slide does draw on ABC perspectives. It's the what and how for the ABC. It's a highle overview of the regulation of the ABC that supports its key asset which is trust. So on the left hand side we have the charter that requires that the services or functions that are delivered by the ABC have certain qualities must be of a high standard. They must inform.
They must entertain and the domestic broadcasting services must be innovative.
the internal governing authority of the ABC as its board and the board has various duties including to maintain the independence and integrity of the ABC to ensure that the gathering and presentation of news and information is accurate and impartial and to ensure that the functions of the corporation are performed efficiently and with the maximum benefit to the people of Australia. So in terms of AI, we can see that the ABC has a core function that includes innovation and the board has a duty to ensure the functions are performed efficiently.
And that's really a clear indicator that we can pursue AI in the context of how to improve content delivery as well as how to improve efficiency in the way that we operate. And of course, we're talking about governance. So we remember that our governance framework must support the delivery of both. It must allow for innovation and ensure that work undertaken with AI complies with these fundamental operational rules and does not undermine the core asset which is trust. Every governance decision we make about AI is ultimately a decision about whether we are pro protecting that trust or eroding it. And of course um just try and move to the next one.
Perfect. Thank you. Um of course opportunities always come with risks. We want our governance framework to include appropriate controls for those risks.
And let's take a look at these before we look at the framework for how we govern the use of AI in a way that manages these risks. We can't of course unfortunately eliminate them. There are six listed but I just want to focus on two of them. Editorial integrity and workforce management because I think from a governance perspective that these the two most important risks.
Editorial integrity. This is the one that has the most potential to damage trust. It goes to the heart of the relationship with the audience and workforce management. We need to equip our workforces with the skills to use AI in a manner that promotes the objectives of public media and respects the safeguards in place and inculcate a deep understanding of the meaning and role of human oversight of AI. We need the humans to manage AI and not the other way round.
I've spoken about the two sides of AI for public media, the efficiency gains and the innovation possibilities. We must ensure that our government frameworks support both.
Next, next slide, please.
Go.
Perfect. I'm just going to talk briefly about the broader legal framework as it can affect our corporate governance of AI and also because last week um Australia did a complete pivot. So on the 15th of July, Australia announced that it would create an office of artificial intelligence reporting directly through to the prime minister's office. And the purpose of this was to allow a consistent and central approach to AI across the country with the work of various different departments that have currently been working on AI being brought together under this central office. The biggest change is the move away from non-mandatory guidance materials to legislation which is proposed to be introduced in early 2027.
There is significant interest from some of the larger AI operators to establish data centers in Australia where as many of you know um there's almost infinite sunshine to power the uh data centers with renewable energy. At the same time, Australia has said that the AI companies will need to comply with standards. They need to earn their social license to operate.
We know that one of the major points of contention with the AI companies is the Australian position on copyright. We don't have a text and data mining exception for AI. And the prime minister has flagged that the work of journalists and creators would be protected. And I've just got one of the grabs up there from him. Australian writers, musicians, artists, and journalists must retain ownership and control of their work.
So, next slide, please.
Perfect. Um, here we'll have a look at an overall corporate governance framework as it relates to AI.
And as you'll note, I've it's the um possibility of extending the existing governance framework. Some organizations will be well advanced in setting their AI frameworks and others less. So either way, you will have existing frameworks that can be extended to govern AI. This is this is an example of one framework.
Your frameworks will depend on the structure of your organization and your governing rules. This is broadly the model at the ABC but not the exact one.
The ABC is a creature of the Australian Parliament. So our rules of operation are quite clearly defined. At the top, we have the board that's setting the AI strategy, a plan for how the organization will use AI to deliver its objectives while managing risks according to its stated risk appetite.
The next level down, we have the executive leadership, the enablement level.
Increasingly, corporations are choosing to have a designated chief AI officer to run everything through one consistent point. At that level is the policy ownership and the coordination of the cross functional response for the ABC. This is where our AI principles sit. Next level down, editorial and operational governance.
This is the layer where you would see specific guard rails for the nature of the AI use reflecting the need to provide guidance to the workforce on the specific use of AI for their work. For the ABC, this is where the usage of AI tools and protection of ABC data standard sits.
Next one's fairly straightforward, legal privacy and compliance. So AI obviously has to comply with the law and any other applicable regulations. And below that the technology management, the management of the AI tools, the approved tool register and vendor management.
Next one please.
This slide gives the governing body the full internal governance stack for an AI use case in one place from strategic intent down to operational monitoring. Monitoring is the practical operation of the previous slide. It's a cascading approach from the board settings at the top to the completion of the review cycle at the bottom. I'll just run through it from top to bottom. So the AI strategy as I mentioned will set the overall direction and objectives for AI use for the organization.
Underneath that we have the AI risk appetite that tells the organization how much risk it can take with AI implementation.
For example, an organization might have a very low risk for inaccuracy in content, but a higher risk tolerance for experimenting with AI to achieve operational efficiencies.
The AI principles have been set to provide guidance across the organization for AI use. After the top three items, we are looking more at implementation of specific projects or use cases. So now we start to look at things like prioritization and re and resourcing is the identification of the proposed use case and prioritization of that use case within the strategy and then data integrity. The data outputs will only be as good as the data that is used by the AI tool.
It's necessary to ensure that data used by the AI tool is accurate and accessible.
It needs to be a fully curated set that excludes outdated material and I'm referring of course to internal use there and the workforce. What does the workforce require to successfully deliver the AI project? Is there sufficient resourcing and training to start understand what they can and cannot do with the AI tool? Impact assessment. Has a legal and regulatory assessment been conducted and monitoring. Once you have started, what monitoring is in place to assess what has gone right and what has gone wrong?
Is there a process to log and review incidents and benefits? I think as one you spoke about this, but um at the time of commencement of the AI project, no doubt benefits were anticipated. They may be f financial or non-financial benefits. Either way, is there a process to review the benefits realization and decide whether to continue?
Then next one. Perfect.
Then these are just 10 suggested governance guardrails for AI within an organization. They're implementable controls, not aspirational principles.
So one, the approved AI tool register.
You cannot govern what you cannot see.
If a tool is not on the register, it's not approved for use.
Define what the human is empowered to do and not do. Be careful about the word review. If you mean approve, i.e. if you mean read, understand, and make a decision rather than just review. This is the single most important control.
Identify where the human oversight and accountability lies and be specific about what is expected of the members of the workforce.
For data input controls, tell staff what sensitive or equivalent material can be submitted to AI tools and again be specific so they know where the guard rails are. Develop AI principles or rules to govern the use of AI and enable delivery of strategy in a manner that's consistent with risk appetite.
Conduct a data governance review to determine whether data is accessible, curated, searchable, and old data is removed or tagged in a way that indicates it's no longer current. This is going to produce far more accurate outputs.
Create an AI incident register. Log errors, near misses, complaints. Report to the board or equivalent oversight body. This is how you demonstrate accountability with the use of I AI and that your framework is working and not just written. Don't forget a privacy impact assessment. You may need one depending on your particular circumstances or applicable laws. And workforce implications.
implement clear guard rails for for the limits of AI use by staff for innovative use cases.
They may warrant an independent review before deployment. So if you're doing something particularly complex or innovative, you might need to think about what extra guard rails to put around that process. And then finally, an annual governance review at board or equivalent level. The technology is changing fast. Continual review of the framework and its effectiveness must keep pace to ensure that the governance remains fit for purpose. And next one.
Great. These are just my personal observations over the last month or so of things that I that have been quite um compelling and have resonated with me.
They're operational lessons that support governance architecture not additional obligations. So the first one is data governance. And the point here is that AI adoption doesn't create a records problem, but it will expose one that already exists. If records aren't well organized, that becomes visible very quickly once someone asks an AI tool to work across them.
>> For example, >> sorry, Jacqueline, you have one minute.
>> One minute. That's on. Um, in terms of inputs, uh, staff need to understand what they can and can't put into the AI.
So you need clear guard rails. And in terms of outputs, it's a really important lesson for staff and important for them to have guard rails. If you don't have the expertise to independently assess whether an AI output is right, you're not in a position to safely use the AI for that purpose. A plausible sounding wrong answer is more dangerous than an obviously wrong one because it won't get caught by a non-expert reviewer.
And finally on the complaint side, just to note that um whilst people are fearful that AI will do away with jobs, I can confidently tell you that um our complaints have increased very significantly over the last year, largely driven, we think, by AI and the use of AI to send in complaints.
So there's an increased workload there.
And then just one final slide, just some takeaways. Um, and the first one's probably just the most important. The use of AI is a governance, trust, and accountability question, not a technology question.
Um, and third, I might just skip to the third one. Humans are probably the weak link. As we see often in cyber incident data, it's the human who clicks on the fishing email. So we need to have frameworks that clearly identify how our workforce can use AI and any limitations that are appropriate for the particular use context. And fourth just just start in terms of getting your governance framework going um and look at how you can extend or supplement your existing framework to support AI and then make sure you continually review because things are changing quickly. Thank you very much.
Thank you Jacqueline. Um certainly interesting perspective you presented.
Now uh let's let's hear from Mr. Alex Raza, director of TV programming RTM.
Okay, very good afternoon. I hope you had a good lunch and ready to actually listen to some more talk. So since morning we've been talking about AI uh which I think for all of us has become uh we are gaining our dependence on it.
It has becomes a tool that we use in our work and um whether it's for AI assisted content or AI generated content.
Okay. So my topic that was given to me is protection of children in AID uh driven media production. But before I get into that, I would just like to uh inform you.
Oops, it's not working.
Next.
Okay. Uh we have an act in Malaysia which is called the employment of children and young persons act 1966.
Uh it's a it's a guide for permitted work operating restrictions legal protections and penalty framework for children and young people. Um next so when we say a child it's for a person under 15 years old and for a young person is between 15 and 18. And um why I'm I'm talking about this is because uh the topic that I'm going to touch on is on um content for uh for young uh AIdriven content for for the young people. And um next please.
Next.
Next.
Okay. But the uh the current act is actually more on um on the physical side because it deals with the machinery act uh the occupational safety and health act and electricity act. It has not touched anything on digital because back that was 1966.
Next. So next. So now that we have AI, so what I'm going to talk about is the digital shield and accountability because um there we got to have a fundamental shift in how we protect children in a modern digital age because for decades the laws focus on physical spaces. You know, factories, studios, and communities. But now the most dangerous media production partners no longer exist in physical studios. They exist in a digital cloud. So uh what I'm trying to propose is is we must shift the burden of child protection away from user and parent parental vigilance and place it firmly onto structural corporate liability.
Next.
Okay. So let's take a look at this uh AI media threat metrics on the on on the upper left we have got deep fakes and cyber bullying which is uh actually the rapid face swapping used to target individual miners among peer groups and on the right AI generated CSAM u synthesizing hyperrealistic exploitative images using non-existent or real children third the voice and identity theft, harvesting voice clips from social media to orchestrate targeted extortion schemes. And finally, algorithmic profiling, predictive engines tracking behavioral micro signals to keep children trapped in addictive loops.
Next, okay, in Malay, in Malaysia's case, this is the evolution of child protection laws.
the first the physical workplace the um children and young people employment act which I touched on just now that it's more on regulating work hours the physical safety and basic labor rules for children and young people then in 2000 the child act was expanded to protect societal well uh welfare focusing on on uh family life rescue mechanism and UNCRC alignment. But today, last year we introduced the uh online security uh safety act 2025 where we enter the digital arena. Here the scope covers things like algorithmic exposure, burden shifts to global tech platforms and statutory penalties that reach up to 10 million ringit.
Next.
Okay. We also have what we call a national child policy vision 2026 to 2030 which is a policy uh from the government which uh focuses on four pillars for children survival protection development and participation.
Survival meaning we want to ensure the 100% legal identity registration and the digital health integration and for protection establishing multi- agency digital dashboards linking law enforcement which is the police regulatory authorities uh we have our commission and social welfare the department of of uh social welfare for real time threat tracking and of course in terms of development We are scaling psychological early intervention programs in schools and the participation to create child le digital safety feedback councils and integrating regional forums like ASEAN ICT forum.
So the core principles in this is the first the best interest of the child family focused approach and non-discrimination and the action plan is on things like cyber security protection against abuse and exploitation focusing on mental health and holistic development and the governance and accountability.
Next.
So we have a four tire cyber response strategy which is prevention centralized identity verification. We have something what we call my digital ID to create verified age gates advocacy school programs teaching digital resilience and grooming recognition intervention rapid respond networks linking hotlines cyber crime divisions and internet uh and ISPs for tracing and finally support specialized psychological rehabilitation for victims of digital abuse.
Next slide please.
So uh when we take a look at this you see on the left the online safety act 2025 short form as ONSA which is uh the legal foundation forcing tech platforms uh into uh 10 million license liability and under the ONSA we have two codes the first is the child protection code and the risk mitigation code. uh the message here is clear that the primary burden is no longer on parents monitoring screen time. It is on tech platforms also to engineer the safety by design environments. So the child protection codes, it's a subsidiary instrument. Uh it was it was just recently introduced on the 1st of June 2026 and uh it forces the digital service providers ones that have 8 million users uh to actually register with the government.
And um it what it says is for people they they won't allow people who are under 18 to register on the um on the apps and um mandatory ID very government ID verification and the platforms have to practice algorithmic detection uh default privacy uh algorithmic and parental and also parental dashboard for parents to actually monitor what the the children are actually watching or consuming online. Next, so 8 million the deemed licency threshold applies automatically to any social media or messaging platform with more than 8 million users in Malaysia.
And um it was not easy actually to get this thing operationalized because there was some resistance from platforms. But we have to think of the children. We have to think of what our people need.
And so we decided that 16 is the age.
And we made it we made them we made it sure that they have to comply with this legally. And uh priority harmful content which is the uh child sexual abuse material and financial scams. And the mandates are immediate dynamic and there is a permanent removal when uh when it's detected that it's the content is harmful.
Finally, the general harmful content, cyber bullying, obscene material and content that disrupts public tranity or accepted cultural behavior because cyber bullying is something that is quite urgent in Malaysia and it's and is trying to protect and uh take care of that.
Next, to protect miners, we mandate a strict age verication path uh pathway. Um they cannot just simply uh rely on self-declared birthday check boxes. The age must be validated against official government records, meaning under 16, absolute block from independent account creation. And for the ages of 16 to 17, account defaults to top tire privacy, harmful media filters are forced on and direct messages from adult strangers are blocked by default.
Next.
So algorithmic filtering and moderation.
Content streaming into feed passes through three mandatory algorithmic filters before reaching a minor screen.
The first filter, the CSEM detection, proactive sweeping and quarantining of explicit material. Filter two, synthetic media labeling, automatic identification and prominence tagging of AI generated manipulated media. Filter three, anti-grooming AI, pattern detection, targeting predatory text behavior and manipulative gamification loops.
Next.
So, uh the risk mitigation code that's the governing architecture. Uh platforms are required to maintain three structural pillars. The first mandatory annual audits certified internal risk teams must audit algorithms and submit safety reports directly to corporate boards. Second, verified advertisers to present scams at placements require governmentissued corporate or individual verification. And third, deep fake identification and infrastructure to automatically label synthetic media.
Next slide, please.
So time is critical when addressing online harm. when priority harms such as grooming on CSAM are flagged for the first hour, mandatory official acknowledgements by the platform and one hour takedown window, complete permanent removal if confirmed and the 24 hours in temporary takedown during extended review and the failure to meet these countdown windows trigger an immediate million ringit fine accuring an additional 100,000 per day until the matter is resolved.
Next, the MCMC legal pipeline. MCMC is the commission for mult multimedia and communications. So, the enforcement follows a clear fourlevel pipeline. The first level, detection and demand for internal algorithmic locks. Level two, voluntary undertaking, which is a binding opportunity for platforms to fix engineering failures. Level three, the 10 million civil debt recovery recoverable directly without requiring a lengthy criminal conviction. And level four, piercing the corporate veil, meaning holding directors and chief compliant officers personally and criminally liable.
So just to give you a brief update um from the 1st of January to the 1st of July uh 345,712 uh messages have been taken down within a 30 to 45 minutes window frame and 91% of these are actually uh gambling ads or scams.
Okay, in order to do the ons uh next slide uh with our legal team and the policy team actually did global benchmarking and comparing with u international counterparts Australia and United Kingdom. So you see Malaysian and Australia has the 16 year old age and uh age verification is mandatory and the primary enforcement mechanism for Malaysia is hourly takedown clocks and civil depths and the maximum primary fines is 10 million for Malaysia Australia $50 million.
>> Sorry Alex, you have one uh >> 18 million or 10% of global turnover.
And my final slide next.
So uh the the AIG the AI governance and ethics guidelines which was developed by two uh two um divisions in Malaysia. One is the ministry of um science technology and innovation and and also the NIO the national AI office. Um they establish core ethical boundaries reliability and control where there's audited parameters for educational and entertainment AI data minimization which is strictly prohibition against logging youth geoloccation or biometrics. The third child ccentric guardrails mandatory shutdown of AI chat bots exposed to self harm or abuse prompts.
Well, at RTM we have actually we are developing guidelines actually on AI in terms of uh to taking uh to to take into considering the social, economic and legal aspects and uh well to conclude my presentation we cannot protect the children of tomorrow using the laws of yesterday. Protecting youth in an AIdriven media landscape requires holding structural code accountable. A child's digital identity must be safe by default and not by chance. Okay, thank you very much.
Thank you, Alex. Another interesting topic.
Next we have Mish Ashwini Natisen who will be speaking about data governance for ethical AI governance. Over to you Ashwini.
By the way, she will be presenting remotely.
Uh, good afternoon everyone. I hope you can hear me.
>> Yes, we can. Go ahead.
>> And I also hope you're able to see the slides.
>> Uh, no, at this moment.
>> Not at this point. Okay, let's try again.
>> Does it work now?
>> Yep, we can.
>> Okay, great. Thank you.
It's a great pleasure to be here and it's also my deep regret that I'm not able to be there in person to make this presentation. I want to thank ABU for this opportunity and particularly Dr. Simantani who's been excellent in uh making sure that I was able to be a part of this. So thank you so much. On that note, I want to go jump into my presentation on data governance for ethical AI governance.
In some aspects, what I want to say is today's discussion slides have overlaps with what we have discussed in other sessions today and I was able to listen to a few of these excellent presentations. But I want to talk about this from the lens of data governance as opposed to only looking at it from a copyright or from a data protection angle. So I'll jump into that. On that note, when it comes to discussions around AI and copyright, there's always an understanding on both the input as well as the output layer.
Since my presentation is going to be on data governance, I'm only going to look at it from the point of view of the input layer. And I'll also be touching upon data governance from the point of view of API open data right to information regimes data protection legislations etc. And to finally draw this linkage of how important data governance is to establish ethical AI governance.
What do I mean by data governance?
Before I jump into the slide, I must say that this is a part of a larger study.
This is a study that was done in seven jurisdictions on data governance in both South Asia and Southeast Asia. What are the countries? You're going to be seeing it in a minute. There was also a mini study on South Korea. I led the study for Sri Lanka and there is a regional report based on the studies conducted on data governance frameworks in all of these countries in South Asia and a few in Southeast Asia. I'm happy to share the report for those who are interested in learning more about what we did with the study. During the study, we looked at what really is data governance. And when we talk about data governance, we are not looking at individual organizations or corporate data management. That is one of the ways to define data governance. But we are looking at it from the point of view of state regulation policies and practices.
And some of these areas that have been included within the scope of data governance relate to right to information, open data, official data, uh access through statistics and census reports, privacy and data protection, cyber security, data localization and of course copyright when as it intersects with data and AI use. Obviously this is a very big report uh spanning to over 100 pages. So we won't have time into look into every aspect of it. I'm going to narrow it down. But when you read the report, you will see that data governance is has been considered from the point of view of increasing access and decreasing access. When I say increasing access, decreasing access, I'm going to speak about those legislations or regulations that increase access to individuals and those that decrease access to individuals. Now I must say the increase and decrease have not been looked at as value judgments. We are not saying that those that decrease access are particularly in poor light. That's not the case at all.
We are looking at it from the point of view of those that decrease access because we ought to decrease access in certain respects. So this is in no way a value judgment.
Let's look at right to information law.
Now when it comes to the right to information law, there have been laws in all of these countries and these are the studied jurisdictions of the study which I said I'll come back to. Philippines, Thailand, Indonesia, Pakistan, Nepal, India and Sri Lanka. I led the study for Sri Lanka. So when we look at the RTI scores, we noted that on paper all of the countries had a right to information legislation. But how those information access laws or the freedom of information laws translated to practice would what was what was the varying factor between these countries. Your question may be how does this relate to AI governance. One of the fundamental principles of AI governance or ethical AI governance is explanability and transparency.
Yes, there can be laws that mandate it and we had an excellent presentation by Mr. Galina who who mentioned about article 50 of the EU AI act on transparency in countries where there is no such legislation. The right to information law can be the mechanism can be the tool through which access to information can be obtained and this access to information can extend to how AI is being used or how AI models have been trained.
On that note on AI and copyright training on AI for training data how AI has been used for training and also copyright protection. So what if training data that has been used is copyrighted material? Can they still be used? I know there were presentations about this earlier today. I won't be going into detail, but this slide will give you a framework about how this has been looked at in different jurisdictions. Most of the countries don't yet have a data or text min or text or data mining exception as should be the case if there should be a free training of data even of copyrighted material but even in the absence of a TDM as we call it text and data mining exception whether fair use exceptions can be used for that purpose that's the next question that comes in but these have not been uh tested tested uh through judicial interpretation through case law. So we don't know yet. There's one case law that's pending in the Delhi High Court in the Indian context and we have similar such examples elsewhere.
But what we need to remember is where AI is being where AI training is being done with copyrighted material. It cannot be done unless either there is an exception or there has to be other justifiable reasons. But right now how companies are operating are because of the legal lacuna. The legal lacuna means that they are able to get away with this because there is no stringent provision on how it can be done or if there is a requirement of licensing etc. This legal lack has enabled companies to train all material available publicly including copyrighted material.
Similarly, the question comes to when it when there is copious amounts of data that are being trained on and such data also includes personal data. Will that be covered within the realm of a data protection legislation? That is the question. Now just because there is a data protection legislation it does not mean that there will be protection for training data that has personal data. That is no reason that depends on certain factors but this is a snapshot just on the existence of a legislation. Is there a legislation or is there no legislation?
But I want to stop here and mention a few points. Just because there is a legislation like I said there is no reason that personal or sensitive data that has been that is available online will be protected through a personal data protection legislation. What it requires is a whether the legislation talks about publicly available data. Is publicly available data included within the scope of personal data protection legislation or is it outside the scope?
Since the country that I work in is Sri Lanka, I'm going to take the Sri Lankan example, the Sri Lankan data protection legislation as an example for this.
When when there is training of vast amounts of data including personal data, what is the scenario under the Sri Lankan law? Just because it is publicly available data, there is no exception to the obligations under the personal data protection act which means that the obligations of consent including notice etc. will apply. Now in the context of a huge data set or when there is going to be generative AI training that's being done this individual level compliance will not be made possible. So the next question is can this be made possible through other ob other lawful basis for processing and usually the answer is is there a particular clause that refers to legitimate uh purposes for processing that is we call it as legitimate processing or business interestbased processing in Sri Lanka. Yes, legitimate reason for processing is one acceptable clause through which businesses are able to use as legitimate interest. So training data could be considered as coming within that exception but it's still not very clear. What we need is clarity on how personal data will be dealt with when personal data is used for AI training purposes. That is another legal lacuna that we have and that we need to fill in. In many countries, for instance, in Japan, they are moving ahead with a clause through which there can be processing of personal data and that processing will be considered as legitimate interests.
So that's something that countries have to consider when they are looking at AI governance and AI governance not just at an individual scale but at the scale of the state as a whole.
So what really is the objective of my presentation? It is to establish that AI governance is not really a new field because we are building it on data governance. There are several aspects that are crucial to data governance and I'm going to focus just on three. The two that we have already discussed data protection legislation, copyright and text and data mining rules and thirdly right to information and freedom of or freedom of information regulations.
Why are these important? When it comes to data protection law, it provides guard rails for personal and sensitive data as may be used when they are being used to train AI models. Secondly, right to information and freedom of information legislations, they provide the impetuous to transparency and explainability.
And finally, copyright because we really need to know how copyright will be considered when the training data includes copyrighted materials. And I believe together we are able to build an AI system that is trustworthy, lawful and locally accountable.
And there are also these tradeoffs that have not really been resolved in the region or in the study jurisdictions.
It's the questions on accountability and secrecy or innovation and protection in resilience and openness. So these are open questions that we are still grappling with but the starting point is to realize that these are the trade-offs that we have to essentially look at.
The other point that uh we wanted to establish through this study was to consider whether countries were learning from each other or they were looking at models in the EU for instance and that's that's been referred to as the Brussels effect. We were not really able to see many examples of how countries were simply copy pasting regulations from others or from the west for instance.
But there are some legislations that mirror the EU example. For for instance, the Sri Lanka data protection legislation models after the GDPR style one. But what we found essentially lacking was the countries learnings learning within themselves. How they were looking at examples in South Asia and Southeast Asia. There were some examples but not so widespread. And that's something I believe that we really need to work on and initiatives like the ones that we are doing through ABU can be the step forward.
So on a wind I'm just going to wind up in the next couple of minutes. But then when we look at the six pillars for rights preserving innovation enabling data governance visav AI governance frameworks these are the points that I want to leave you with. And what are my practical takeaways?
Essentially the first question is where does your data sit?
Secondly to push for clear text and data mining regulations. Third is to consider regional coalitions and not just ones at a national level. And the last point being transparency really is the leverage. And I'm glad to see that how presentations throughout the day went into great detail about how essential it is. On a concluding note, instead of leaving you with just answers, I believe that these are the points that we can all ponder. Where really does our national data governance framework help?
Is it blocking any innovation that we may have or is it really enabling responsible AI use including when it comes to broadcasting of content? How are we looking at copyright exceptions and is it keeping pace with our requirements of today including openness and also ones that need to be privacy respecting. And finally, what is the role of ABU like regional organizations when it looks at AI governance as a baseline and how do how can we look at it and how can we consider them. So on that note, I want to stop my presentation. I don't want to go beyond the time that's allocated to me. But I hope there will be questions and I hope that we can engage further on this. And once again, apologies for not being there in person, but I've put my LinkedIn ID here and I'd love to connect with any of you if you'd like to chat more on this. Thank you so much.
>> Thank you, Ashwini. What uh I really like your last sentence there. Can you repeat that your last sentence there? What would be something to do with Abu?
Can I see it again? Your last uh let me What would a regional ABUled data and AI governance baseline look like? That is a very very powerful question.
So we have approximately about 15 minutes for Q&A. So I would like to open the floor for any questions. Anyone?
or I should start myself.
This is for Jacqueline since you're talking about governance like for any broadcasters like um that are starting their AI journey right for example like um smaller broadcasters with very very limited budget. So what will be uh the two or three elements that they should implement first?
Thank you. Um I think as I as I was talking about most of us will have existing frameworks and I think they can be extended. So starting with what you've already got I think makes sense and then seeing where you can add on to cater for the particular AI enablement that you're interested in doing.
And I think um secondly would be building the st the um the staff capability to manage AI to to understand how to use it, understand how it can be used to improve efficiencies and understand the guard rails so so that they can understand um the limits on what they can do with it in the workplace. But I think the key thing is to just start and work with what you've got because we're all I think it's an exciting ride for all of us and we all just um >> refine it as we go as you see.
>> I think we are actually because it's happening all around us. So we need to just um use the frameworks we have and extend them as we can and as makes sense and suitable, appropriate and ethical.
Uh, next question. Anyone?
Well, I should ask again.
No. From the floor.
Good.
So this is a kind of a very general question to all of you and because now there are you know Alex from Malaysia and Jackie from Australia and of course you know as one of Simantani from Malaysia as well. Um Alex uh made a presentation about a kind of you know uh measures to protect you know children younger generations from uh well kind of a misuse of AI I think and um this movement is actually sort of spread out universally uh like when we think about Europe for example and Alex presentation uh showed the case of UK as well but not only UK but also other European countries as well like if I remember correctly um France or Spain. Uh do you think this movement is going to spread out because well um and um I also hear that well this is kind of ideal for in terms of the regulation but when it when it comes to the operations of practices for example of course know people will be able to tell a lie about the ages for example and Malaysia will be probably they're trying to well probably make sure that uh they will um connect uh with the data about age uh with the verification end uh do you think this kind of measure is sort of uh well do you think it is going to work um in general >> okay thank you Harusan okay we just started that in June so for me to say whether it's going to work or not I'm positive that it will work but how far it will work I think remains to be seen because I think we can see maybe in 3 months time how how many more takedowns and on on what basis they'll be taken down and maybe in in another 6 months and in one year uh because I think every country has its own sovereignity and every country although we would say these days that children are very generalized but I think the needs and the uh dynamics of each country is still different. What may be the needs of Malaysian society may differ from Japanese society maybe or African societies or European societies or or American societies.
So I think uh primarily we have to start somewhere and that was uh the first start for Malaysia to tackle the whole issue because you know you cannot stop technology AI will definitely develop.
We used to talk about generative AI but now it has advanced to another level where it can start thinking for itself and we don't know what's going to happen maybe next year or even in the next month but we have to make uh a start somewhere because if not we'll just be thinking we'll be talking discourses we'll be talking hypothetically without taking any actions and of course plans of action that we take although we usually generalize them into uh short-term middle-term and long terms we still have to be amended and be responsive to the changes in society to changes in the dynamics of the world. For example, um let's say you you have this identity thing, right? Although that's one attempt for Malaysian government to have identification, but I'm not sure if all countries in the world require things like identification cards. like Malaysia, yes, we have our IDs, but I'm I'm not sure how many other countries still practice that. And so that's also a factor that we have to take. It's how to identify and how to identify that uh the person who's actually attempting to go online or on board the platform is that particular person or somebody impersonating and now you know you have the use of VPNs that can actually deter people even in places. So things are developing and that's why we have uh the ministry of science technology innovation we have the national AI office and several other governing bodies in the country to look into things and address them as swift as possible. And of course when we talk about laws or guidelines it's not a matter of one one particular um what uh agency you have to have public engagements because there may be people who may talk about for example the freedom of speech which is one issue when when we had that and the right to information as was um uh that was asked by uh Ashwin just now. So things like this are always being considered but I think whatever it is we have to take the make the first step. If it's a baby step at least it is a step like you say like people say this a journey begins with a single step. So that's one thing that uh we we're going into and we are actually uh considering what other people are doing because we don't want to be isolated by ourselves because kids will grow up information is at the fingertips and even though there's parental uh thresholds when the when the kids grow up they have their freedom how do you control them so things like uh getting them cultured through responsible use of AI and knowledge So that has to be uh has to begin.
>> Next question. Yeah, we got time.
>> Hello Aswan.
I've been listening to this from the morning and also uh what uh the ABC perspective also from RCM two of the bigger players in the region uh that we deal with and coming back to you as one in your presentation.
It's not a question but it's just a thought to get people probably thinking probably these other newsrooms etc that are smaller. Is it time that your yearly budgets that you prepare includes this aspect of what we've been discussing? Because we can talk about it and what Alex mentioned is very true. We have to it's it's there. We you can't deny it. You can't avoid it. You can't probably, you know, ignore it. But from a financial aspect from a financial point of view probably it is time that our yearly if it's not already being done by most is have budgetary allocation to be able to achieve these things or protect our broadcasters individual broadcasters or newsrooms or content creators within the circle. I mean that's that's what I think. I could be wrong. uh the finance expert is there but uh I think if we don't we probably will end up plucking funds from other aspects of the business and trying to achieve something which in fact is also impacting that other areas of the business just just something that came to mind. Thank you.
>> From my perspect very interesting Indra, thank you for the interesting question.
From my perspective, we should include it because at the end of the day, we have to go back to the basic what are we trying to achieve?
So once we are clear on what we want to achieve, then comes the question, how are we going to achieve it? So your question I think comes from in the how.
So because what management or the stakeholders wants to see is basically the end figure. So how do we see the end figure? We need to we need to have all the right information. Correct? So if we are reporting a half big reports for example. So if we are not including any certain expenditure that we think of including so we won't be able to actually you know find it.
>> Sorry I was just to add I'm not only talking new AI tools that are coming on in the market. I'm talking about legal.
I'm talking about you know training needs legal side uh AI side all that inclusive which too often I've seen um u whereby companies just totally disregarded then all of a sudden in the seventh month of the year you are running to the financial controller asking him for some money.
Yeah, that happens like 90% of the time and it requires special special requests, isn't it? But it's always good to plan. It's always good to plan. And I always believe that if you you fail to plan, you're planning to fail. So that's my view on it. Unless the the rest you have anything else to say.
>> Yeah, I do agree. Do we have to plan and I do agree budget should be allocated for that but you know when you talk about national budget in Malaysia for example uh we get our budget tabulation in October so sometimes we don't know what the prime minister has to say so on that particular day on that Friday in October we will listen and see where he's focusing and AI is something that he's very serious about Malaysia for example that's why originally AI was under uh the ministry of science science and also the ministry of communications and multimedia and then it was divided that uh it became its own ministry the ministry of digital and uh the national AI office the na that I mentioned just is actually under the ministry of digital and they look into things and in terms of budget when you're a government staff you have your um government um email for example which is under Google you you get the facility for using Gemini AI and then for example if you go training under let's say a university I went for training under one university in Kalumpur and what they did was when you became a student even though it was only for 3 days they gave you a free professional account for chat GPT meaning they actually um getting you used to the idea that AI is something that you need to do but they will teach you how to use it uh responsibly For example, now these days if you look at uh if I look at my friends, they enjoy using AI but to make posters to make funny videos and things like that which for us who are content creators and we are serving the public that may not be something that we should strive for. But you should use um um AR responsibility which I think most of us have used in our presentations from boarding uh because we want to find out how to do things and I think that's something that is important because sometimes certain vocabularies don't appear in our minds for example uh I'm a producer and uh basically uh IPLC is more on the legal side but of course we have a certain knowledge of legal but I may not be uh as fluent as Dr. Sema for example, she's been there for so long and she that's been a study. So using AI is something important and I think our government is very serious on that and that's why they are exploring and they're finding new things uh and at the back of their mind knowing that it has to be sustainable for the people and yet has to be agile and flexible to accept new things coming in.
Indra, can you pass the mic behind?
We have one more question.
Thank you for the opportunity. Uh I have a question for the for ABU. So uh TVRI has been a member of ABU since 1983 and we are one of the seven founding members in Asia Vision since 1984. So uh my question is uh should AU develop a road map policy for AI for its members so that we can have a uh road map for all the ABU members.
Thank you.
Thank you. Um I think the technology department at the ABU has recently launched an AI uh site for the members.
It's a work in progress. It's not perfect but uh I know when Dr. Vasel was around director of technology he had initiated that. So there's a site which we can share with you and of course input with the likes of yourselves and the other members will help us better.
But it's an idea that came I think some years ago and it's finally started uh to it took off and it's there now. Thank you.
May I be cheeky and ask a question since uh our colleague from TVRI has talked about road mapap for a uh concerning AI right and as I mentioned our government has actually sponsored us using Gemini and um chat GPT will there be anything in the future planning maybe Mr. Aswan um Dr. Sema uh or even Indra that uh ABU can provide some kind of um facility for AI for members.
I can speak from the news perspective because the other departments I'm not so sure but uh we have been talking about it because and I think it's not only about established members I think the smaller members as well that we have that I keep on mentioning about there is a need so from the news perspective yes I can give you a little win that I had when I started a year ago is we managed to convince to get Gemini for our newsroom team and this is because when I walked into the office the first question was what tools do we use we didn't have any tools at that time not a paid version anyway so it was on the website whatever it was so now we've started doing that to help in translations which is of course very much specified if we use it it's translated by AI similar to EBU so from the news perspective yes from the whole ABU perspective probably at the GA you'll be able to ask That's why I say I'm just being cheeky.
So the ABU has launched its own AI guidelines and these are not binding but uh you know they serve as best practices for members and for uh ABU to come up with any binding guidelines uh you know it would really depend on the willingness of the members because coming up with something binding requires a consensus.
>> Yeah. And I don't think if it's a binding thing, it would actually work well with the governments because I think every country will want to practice its own sovereignity. And I think primarily we will have to follow the road map set by our leaders rather than but of course having a a guideline is good because that would be useful for broadcasters because I think the road maps developed by governments are very general but uh it's great to hear that uh you're developing the guidelines for broadcasters.
interesting stuff like yeah okay I think time's up for session three and uh thank you Jacqueline Alex Ashwini and Simma for requesting me to moderate this session Hello everyone. Uh this is Estra from TRT Turkish Radio Television Cooperation.
Uh it's my pleasure to moderate uh this session. Thank you so much for uh this invitation for me. Erh I am only one on the stage uh because our uh speakers uh attended it will attend us online. Uh so I will start.
Uh welcome to session four. Uh it's uh our title is global perspectives on AI digital media and emerging regulatory challenges.
It's uh today we have experts uh from international organizations, academia and public broadcasting institutions.
If you are ready uh we can start uh because I guess our speakers on uh their screen and they are waiting us. Uh first of all I will start with Dr. Mary Green. Uh our presentation is uh how can platform regulation and enforcement trends across Asia Pacific. Uh I would like to tell something about uh Dr. Green. He is international media law advisor and also academic with extens extensive experience in platform regulation and communications law across Asia Pacific.
I'm not sure is he ready or not.
>> Ezra, thank you very much. Can you hear me?
>> Yes, we can hear you, Mr. Welcome.
Welcome to our session.
>> Thank you. Thank you. Thank you very much for your uh kind introduction and greetings from Melbourne where it's winter and it's very dark. So, uh the sun is still shining I think in wonderful Jakarta.
Uh even though remotely, it's a great pleasure to be back in the company of Radio Republic Indonesia uh and the ABU IPLC. I've got very fond memories of working with former RRI director Pani Hardy. I traveled with him in his business car and got a very quick introduction to Jakarta traffic.
Uh Pney was such a practical person. His uh front passenger seat sprung back to lay flat and he very adroitly positioned himself uh in a way that he could have a nap on the very substantial journey to his next appointment.
RRI has always been very resourceful and uh Pney certainly led the way in relation to that and the IPLC also has had some ongoing and very significant impact on member states as we wrestle with the regulatory impact in a fast changing media environment.
And today what I want to do is to note the nature of that change and identify some regulatory initiatives in the Asia-Pacific that are shaping major reform in media accountability.
Our focus will be on the shift in audiences and audience behavior from freeto-air broadcast media to online delivery.
Now I've got a PowerPoint but I won't share it with you because it'll it'll be available from the secretariat uh later in the conference uh and it just it just gives you the references that I'm going to allude to in my brief comments.
Over 10 years ago I undertook a study for UNESCO into the development options for public service broadcasting and community broadcasting in the Mikong.
And the report looked at media environments in Cambodia, Lao, Thailand, and Vietnam. And Singapore was the reference point uh by way of contrast.
There we I see the the PowerPoint. So if we can just flick through to slide number two, that'd be great.
Um, and uh, one of the things that became very clear when I was doing the study was the durability of the 2008 UNESCO media development indicators, the MDIs.
Uh, and they had four core category areas. Now, if whoever's manipulating the slides for you, maybe I can do it myself. No. Um anyway, these these are all outlined in the in in the slides. Um in slide two, uh there are five key areas that the MDIs uh looked at and addressed. First was system of regulation uh to look at laws whether they in fact support uh regulators being independent and having the ability to foster freedom of expression. Number two was plurality and diversity of media uh and to see if public the public has access to many types of news sources particularly three media as a platform for democratic discourse to ensure all people have a voice in society and that uh news and the presentation of news is fair. The fourth element in these MDIs was professional capacity building to make sure journalists were safe and had standards of presentation and news gathering. And finally, fifthly, a technical and infrastructure capacity where media organizations had the ability to adopt uh new initiatives.
So these five MDIs even though created uh in 2008 seems a long time ago uh I think uh they have some enduring significance in the comments I want to make.
Now this is really worth looking at in terms of the way we are facing quite a sea change and it's forcefully demonstrated in the changing ways in which audiences are now accessing content and structurally this is evidenced by a separation a growing separation between content makers and content contri distributors. is so the split between who makes the content and who delivers the content used to be the same entity. Now there are different entities in different parts of the world often.
So what sort of pressure does this place on media regulation and media regulators among the foremost of current questions is that of regulatory jurisdiction.
If content is produced in one nation and distributed to that nation's audiences from another nation, how far is the reach of a media regulator?
Let us look at the gravity of this challenge by examining the production and distribution of news. And I want to share with you some Australia Australian work uh on this matter.
Uh PowerPoint number three is relevant here for those who've got access to it.
The University of Canberra UC recently published the digital news report for 2026.
And this work is part of Oxford University's Reuters Institute for the Study of Journalism and their research that examines news content origination and distribution in some 48 countries across six continents.
The UC study found that social media is now the second most Australian news source behind television.
Social media 56% free to air television 57%. There's nothing in it. So social media has even become the main pathway for online news websites. Some 32% of people are now getting their news from not newspapers or watching on screens in terms of free to air but are getting it off websites and even access to those websites is now coming through social media platforms.
A pressing regulatory and economic issue is who creates and bears the cost of news content and who benefits from any income associated with it. On to PowerPoint four. Now, there have been two major responses to this question in Australia.
The News Media Bargaining Code, the NMBC, was established in 2021 under the watch of the Australian Competition Regulator, the Australian Competition and Consumer Commission, the A C. Now, this code governs the commercial relationship between Australian news businesses and the designated digital platforms who benefit from significant power imbalance.
By December 2021, over 30 commercial agreements between digital platforms, principally Google and Meta, and Australian news businesses had been completed. And over the five years, the past five years, some 1 billion Australian dollars have been paid by the digital platform to the news creators.
Now, when it came to renew these relationships, Meta decided that they would not continue the arrangement and removed or deprioritized Australian news in its platforms. And so just recently the Australian federal government has responded with a new code and a new provision, the 2026 media bargaining initiative, the MBI.
And this provides that larger digital platforms will now pay a levy of up to 2.25% of their Australian revenue unless they come to a deal with local journalism enterprises.
Now, this arrangement is expected to generate something like between 200 and $250 million uh a year that would be distributed to local news providers.
Now, you can understand that the reaction to this development uh created some hostile reaction from some of the players. A meta a meta spokeswoman said this initiative amounted to what was really a digital services tax.
However, under the MBI scheme, the rate of levy can be reduced if individual commercial agreements can be reached with news providers. So, if they can do the deal and it's acceptable between the news providers and the digital platforms, then the levy will not apply.
Now the N MBC and the MBI test was really a stretching exercise for the regulatory reach of national governments over the businesses of international digital providers who collect the revenue from content generated by national providers who bear the cost of production.
that the NMBC worked at least for a while may suggest that with further consultation there may be an effective way ahead uh and news organizations are optimistic that this can be achieved.
Now, online engagement does not only raise questions about the disconnect between who produces and who distributes and appropriate means of cost and revenue sharing, but also concerns itself with online safety. Now, we've had some very interesting discussions so far in this gathering today that I've been able to hear part of and uh I'm very impressed by the insights.
Now, Australia, PowerPoint number five, Australia is now a leader or was a leader in and remains a leader in legislatively engaging with child safety online. We've had a lot of discussion about this in the in the in the short time that I've been with you this afternoon.
In December 2025, children under 16 were excluded by law from holding social media accounts.
Let me say that again. Children under 16 excluded from holding social media accounts.
Companies such as Facebook, Instagram, Tik Tok, YouTube X and Snapchat are required to put in place mechanisms to ascertain the age of users under 16 and prevent their access.
Now, if the social media company doesn't get involved seriously in making this happen, they could face fines up to 49 million Australian dollars. Now that 49 million is a large amount but has just recently been raised to 99 million. So there's no doubt about government resolve.
But however the effectiveness of these measures is open to question really. A study by the University of Newcastle in Australia found that more than 80% of under 16s in the country were still accessing social media platforms after this requirement was put in place.
Much of the verification was self-generated. You just said how old you were and that was it. Uh and that might be matched by the upload uploading of a of a face image which would be tested for for age.
Now this is requiring some further scrutiny. But the Australian initiative, which was a world first, has created a lot of international interest.
And in our own region of the Asia-Pacific, Indonesia, Malaysia, and Singapore are among the nations, as we've heard today, that have uh engaged with the reform in children's online safety.
Indonesia was the first country in Southeast Asia to restrict children under 16 gaining access to social media.
And that prohibition came into effect beginning from March this year and is matched by evidence of some regulatory resolve. Uh the Indonesian Ministry of Communication and Digital Affairs undertook a a surprise visit to the Jakarta Office of Meta recently uh following concerns about harmful content and how it was managed.
But the scale of the task in Indonesia to manage uh around 70 million users, that's the 25% of the population under 16 is immense. But there's no doubt about the resolve. Communications Minister Mata Huffford recently told a news conference that there would be no compromise on the question of compliance.
Now, as we've heard today, Malaysia has also developed explicit requirements into age verification to prevent under 16s accessing social media. Amendments to the online safety act effective from 1st of June this year center around rigorous means of age verification.
Malaysians are far ahead of the Australians when it comes to this matter. My digital ID or passports are means that are provided and only accepted as a a a proof of identity and age.
The Australians need to have a look at this.
Singapore has taken a different approach with an emphasis on a differentiation of safety standards.
Uh recently digital development and information minister Joseph Tio said that the same services can provide safe access uh for those under 18 while some services may not do that. And so rather than having a blanket approach the Singaporeans are looking at developing differentiation.
restricting access is not where they want to start and and not the preferred outcome.
Now, there appears to be growing support for regulating social media use by young people and the efficacy of verification methods of age and any substantial engagement by digital platforms uh appear to be remaining issues to be developed and resolved.
This area of regulatory reform though is indicative of the challenge of jurisdictional reach when major offshare corporations own and control national media use.
We have just touched on the increasingly demanding issue of young person access and I want to look forward to particularly hearing what Dr. Ammo Panchiwa has got to say after uh my presentation and in in our session later today.
And in closing, I just want to note some developments that are pressing regulatory issues that face nations of different scale in the Pacific.
A principal resource in looking at the resources of Pacific Media uh is the state of media report uh Pacific 2025 published by the excellent work of the ABC the Australian Broadcasting Corporation's international development group.
12 different countries in the Pacific were studied. The Federated States of Micronesia, FSM, Fiji, Kurabas, the Republic of Marshall Islands, Naru, Nui, Palao, Samoa, Solomon Islands, Tonga, Tavalu, and Vanuatu. I've worked in many of those places and they're all great nations, but like any small nation or smaller nations, they have particular challenges.
And the findings of regulatory significance in the Pacific uh include what are patterns that we've observed elsewhere international international internet access dramatically increasing particularly in Fiji and Samoa uh with internet access rates now of 85 and 75% of the nation respectively.
misinformation and disinformation as being major issues in terms of media content.
Government funding of media inter enter enterprises both sustains and stifles media in some ex examples uh and media freedom is a concern of many Pacific media institutions and media selfcensorship tends to be a characteristic of some of the smaller Pacific nations where delightfully everyone knows everyone and so when you are running news stories about your neighbors or your friends. Uh there are different criteria often that kick in.
The extent of government constraint of the media was amply demonstrated by Fiji's media industry and development act the MIDA which was introduced in 2010 and repealed in 2023.
Now this original act was established by decree and then uh military intervention uh by way of the military government and this significantly restricted media freedom uh and stipulated that almost all Fiji media had to be locally owned and there were jail terms for those who reported or were judged to be reporting uh content not in the national interest.
The repeal of the MA has now put in place an environment of self-regulation under the watch of the independent Fiji media council in Fiji and media freedom has now been uh effectively restored.
What I've attempted to do, this is the last PowerPoint 7. What I've attempted at tended to do in these comments on platform regulation and enforcement trends across the Asia Pacific is to identify some key developments and the a Asia- Pacific has undergone rapid change in media use and my earlier UNESCO study seems a long way away but the issue that we look issues we were looking at developing public and community broadcasting um still have the same sort of challenges that we've outlined in our brief survey this afternoon.
But the media development indicators which informed much of that study still remain very durable.
What is central to my comments today in conclusion is that media regulation must follow where the audience goes and this can take us into some very demanding places including testing our understanding of jurisdiction and when there is a split between who the content providers are and who are the content distributors. This split is getting wider and a big challenge for all of us as we chart the future.
Thank you uh to the moderators Ezra and our hosts the impressive RRI and our organizations our organizers the ABU IPLC under the great leadership of Mr. Haryuki and our go-to person of expertise Dr. Simantani. Thank you very much.
Sorry. Thank you so much Mr. Green. Uh your presentation is really sign significant to understand especially Asian Pacific perspective for us. Uh please wait us. I guess our uh audience wants to ask some questions to you.
Thank you again. Thanks for your time.
And next time uh next our speaker are they are too uh because of they are from VIPO everyone knows world intellectual property organization Michelle Woods and Yasharia. Their uh topic is harmon harmonization of AI copyright rights protection frameworks.
If you are ready, sorry, I can say again. Uh, sorry for I I want to say again their topic. Sorry for this. Uh, their topic is uh copyright law division.
Uh, Michelle W and Yasharia. If they are ready, we can start.
Hello, this is Michelle and I just wanted to say hello to everyone but in fact Yash will be presenting for us today and then I will be available to join him in answer any questions you may have.
So should go over to Yash who has our PowerPoint as well.
Uh thank you Michelle. Uh can we have our presentation like slides please?
I'll start.
I think it it's lagging but it's fine.
Uh good morning and good afternoon to all the panelists and the participants of the session. Our presentation is an update from the secretariat of VIP post standing committee on copyright and related rights or the SECR on the committee's work concerning copyright and generative artificial intelligence.
So I'll briefly explain where the issue sits within the SECR, how the four information sessions that has happened previously have developed the scoping study on copyright and generative AI requested by the committee and the next steps which have been identified for SECR 49.
If we can go to the next slide. Is that possible?
Yeah.
So where AI sits in the SECR work program.
So within the SECR the work on copyright and generative AI has developed under the agenda item other matters in the broader discussion on copyright in the digital environment. The committee is the intergovernmental forum in which WIPO member states discuss substantive questions of copyright and related and the secretariat's role is to implement the mandates given by the committee and to facilitate the discussions. The process shown here has four elements.
Information exchange, comparative updates on legal policy and market developments, a more structured factual basis through the scoping study and the discussion by member states.
Can we go to the next slide?
So as we can see here the is the timeline of how SECR developed. Uh the timeline shows the work has developed in response to success successive committee decisions. During SECR 44 in 2023 group B proposed an information session. The proposal expressly described the exercise as an exchange of experiences and perspective rather than a norm setting exercise. Then successfully through SECR 45, 46, 47 and 48 we had different topics, different sessions. Uh and the topics were like for the first session we have impact of generative AI.
We were discussing on inputs and outputs. Second session was about comparative jurisdictional approaches to training, copyrightability etc. Third session was about visibility, transparency, remineration, licensing and we also got the mandate for a scoping study. And the fourth session, the latest one was a member statesled initiative where we discuss about case laws and the recent developments. And I will be discussing about them in greater detail in the following slides. Next slide please.
So what the SECR secretariat has delivered? The secretariat's contribution is not only limited to organizing panels. It includes consultation, curation, convening and documentation. Four information sessions have been held so far. All the information sessions are in six languages and we also have supporting materials made available in all those languages through VIPOS documentation documentation system and with a public record through documents, speaker profiles and webcast. So all of these sessions can be find in the VIP post webcast and on the SECR page we can find all the documents related to the AI info session and other agent items at the SECR. Uh the programs have also included four perspective groups such as governments, creators and right holders, creative industry peoples, AI developers and platforms, legals and technical experts and also observers. And the whole purpose is to enable member states to hear different experiences and approaches in a structured setting and to preserve a discussion. Can you go to the next slide please?
So now we discuss in detail about what were the discussions at SECR45 which I call as establishing the landscape. The first information discussed the emerging landscape. uh it examined the growing use of AI in the content creation and its practical effects on creators, business and users across different creative sectors. It considered both opportunities and challenges including the relationship between human created content and generated material. The second panel looked at enabling factors for beneficial use. This introduced the distinction between inputs such as the use of protected content in machine learning and outputs including the legal treatment of generated material. It also covered technical standards, licensing modalities and possible new sources of revenue. Next slide please.
At SECR 46 there there were again two panels. This format was followed until SECR 47. And in the first panel uh there was this formation exchange became more comparative and more legal in focus. The first panel considered copyright and AI training uh which rights may be implicated, how different interests are addressed and whether exceptions or limitations may apply in particular jurisdictions. The second panel addressed AI generated and AI assisted material including authorship, copyrightability, infringement, liability and any specific legal or procedural requirements. For example, uh the Republic of Korea during that session described a government working group that had produced guidance and organized separate work stream on AI training and AI generated outputs illustrating one consultative national approach and previous like slide seven.
Yes, thank you. So slide seven uh sorry SECR 47 also talked about visibility, accountability and participation in value. Again it followed the same format of having the two panels and the discuss and the discussion was confined to two operational themes. First on visibility and transparency which was followed by licensing and remineration. The first panel considered tools such as data set provenence, labeling, traceability, rights management information and technical measures. These tools were discussed in relation to making the use of origin of content more visible. And the second panel considered licensing practices, remineration mechanisms, data related questions and technical measures that are being explored by different stakeholders.
Can we go to the next slide please?
Now this was the last and the latest information session. Um in this session we had three panels and it it was designed as the current state update. Uh so first panel was with member states presented recent national or regional initiatives including studies, consultations, guidance, legislative or regulatory developments and policy proposals. In the second we had experts who provided an overview of developing case laws across jurisdiction. This reflected the growing role of courts in addressing questions concerning training, evidence, transparency, outputs and liability. The third was we had representatives of creative industries who described contractual tools and practical experience concerning protected content used by AI system. As the result, the committee welcomed the session and agreed that the next meeting should combine discussions of the scoping study that was approved in SECR47 with a further information sharing session between member states.
Can we go to the next slide please?
So the new SECR mandate is a scoping study on AI training and copyright which was a significant development at SECR 47. The mandate asks the secretariat to prepare a study on policy or regulatory approaches to the relationship between AI training and copyright together with related rules and applicable practices concerning authorization, enforcement and compensation for use. The study should therefore be presented as a factual and comparative resource requested by the committee and not as a recommendation for a particular regulatory outcome.
Can you go to the next slide please?
So this slide tells about the emphasis scope of the study. Um the first part it is intended to explain what happens in the training of generative AI models including the terminology the use of protected content during training and what occurs after the initial training stage. The second part is intended to examine licensing and compensation mechanism currently found in different member state including how training data is obtained, the rights that may be implicated and the voluntary collective or other relevant mechanisms. The third part will survey important case laws and practical litigation issues in an evolving in this evolving field. Can we go to the last slide please?
Uh to conclude last thank you to conclude the committee has now supplemented the information session process with a scoping study which will be presented and discussed in the next SECR 49 currently scheduled for 30th of November to 4th of December 2026 and the secretariat will continue to facilitate the process in accordance with the mandates and directions given by the member state. Thank you very much. We also have these QR codes uh and you can directly reach to the webcast by scanning them.
Thank you.
I think so. Thank you.
>> Okay. Yeah, it works. Thank you so much, uh, Mr. Woods and, uh, Miss Arya. uh this is really important for all of us and uh we are happy to hearing uh detailed information from uh experts like you. Thank you so much.
>> Yes, I guess last version. Yeah.
>> Yes. Uh we are on the last stop uh from Ebu Mik Evan Galishta. I hope uh I say your name correctly and he is successful senior legal council from European Broadcasting Union. EU everyone knows and his presentation is AI and copyright in the European Union.
And thank you so much if you are ready we can start. Okay. Hi HRA and good afternoon everyone. Thank you to Seantani and the old ABU for inviting me today.
Um so my presentation will be about the um I will give you a framework of the uh EU legislation concerning AI and copyright and what the advocacy position of the view on this subject.
So when an AI system wants to scrape your content to feed the training data uh of its system in principles based on copyright law, it requires the authorization of the rights holders.
However, in the European Union, we have introduced in 2019 an exception for text and data mining that allows the reproduction and extraction of content for text and data mining for scientific per uh research and uh commercial purposes.
In this second case, when the uh scraping is carried out for commercial purposes, your legislature has provided the rights holders with the possibility to to opt out to reserve their rights and claim back their copyright. The opt out must be declared in an appropriate manner which when the content is freely available online has to be machine readable.
We have seen over the past years various methods uh that the right holders have adopted to express this opt out. One is to make a statement in the um terms and conditions of the website. Here in this slide we have an example of one of our members uh that stated that the content hosted on its website cannot be used to train artificial intelligence systems.
But probably the most um widespread method is to record to the um robot.ext file of your website. This is machine readable because it communicates directly with the crawler of the AI systems.
This method however is not very efficient because as we say in the DBU it's a flag it's not a shield. So it communicates the crawler that is not allowed to scrape content from the website but it doesn't actually prevent the crawler from doing so. There is also another concern uh that we call the granularity issue uh which regards the crawlers of the search and engine platforms such as Google. uh we are worried that by blocking the crawler for AI training we also block the same crawler that index your our website in the search engine results and the last issue concerning this method is uh that uh applies only on the website level. it does not follow the content when this is shared on outside of the website such as on thirdparty uh platforms like social media. And this brings me to the third method which is a contentbased standards uh where the opt out is declared is embedded in the file of the content. We have various protocols uh such as the C2PA uh the content coalition for provenence on authenticity which is widespread in the audio visual industry. Um the advantage of this method besides being content based is also that you can embed other information besides the opt out such as licensing conditions to exploit the content.
But um the problem of these contentbased standards is that often they are sectoral or limited to a a specific type of content. For instance, certain standards apply only to text whereas others like the C2PA is very uh easy to apply for audio and other visual content.
Now given this uh panorama in which we have a a fragmented use of the opt out declaration, the European Union has published recently just two weeks ago a study in which it proposes uh the creation of a registry uh to record the opt out of the rights holders. This registry is not meant to replace the methods that I have listed before but only to complement the protocols that right should holders have already used.
Um the registry is uh based on a digital fingerprinted technologies that recognize all different types of standards uh and it's based on an hybrid architecture. It's both a central resolution layer and relies at the same time on distributed on boarding nodes.
Although we appreciate this initiative of the European Commission, uh, DBU and many other stakeholders uh, in the creative industries of the European Union have certain reservations concerning this uh, this proposal uh, especially with respect to its efficiency and the cost associated with the building of such a system.
Now we move to the advocacy activity of the view concerning AI and copyright. Um we have two proposals. One concerns the licensing of content by broadcasting organizations for AI training. Given the business uh model of broadcasters, we distribute cont programs that contain both our content for which we hold the copyright and third parties content for which we only acquired a license to distribute.
Um so this prevents us to license our programs to AI systems for AI training because we will need to seek for the authorization of the so-called underlying rights holders. So owners of third party content. Um, so we propose to reproduce the cable retransmission regime that has been in place in the European Union for over 30 years where when the a cable operator wants to retransmit the broadcasting signal of a broadcasting organization, it he needs to conclude an individual license with a broadcaster on the one hand and clear the underlying rights through collecting management of rights. In other words, by paying collecting societies.
We we think that this approach that draws from the benefits of both the individual licensing and the collective management of rights can enable broadcasting organizations to license the content and negotiate with uh AI systems.
Then we also propose to complement the uh existing legislative framework of DU by introducing a right to remuneration that would supplement the text and data mining exception that I described at the beginning of my presentation.
In this case, we would uh kind of take inspiration from the uh private copies levy system uh where if a right holders has not declared the opt out and therefore makes available his content for ice scraping, he will be entitled to a right to marination whenever an AI system has scraped his content uh to feed his uh training data.
And uh this brings me to the end of my presentation and I thank you everyone for your attention.
>> Thank you so much Mr. Evangishta. This is really important topic and hot topic also text data mining exception. Thank you so much again. uh our speakers please wait us because our audience can uh maybe ask some questions to you because all of yours presentations are really significant and uh important for us. Thank you so much all of you and today's last presentation is uh from Dr. Amal and uh independent broadcasting consultants and internationally respected broadcast engineer. Uh we are I guess uh we have presentation and now maybe we can start. Uh this topic is protection of children in digital and social media ecosystems. It is a really interesting topic because children are really important and we need to protect them in everywhere. So uh per uh time uh peer prayer presentation also we had a chance to hear something about uh protecting protection of children. So looking forward to hearing something from Dr. Amal.
I'mma greetings to you all from New Zealand.
Outline of my presentation. I'll give a quick background to the topic uh in in the uh in this uh conference under the team and the the harm due to the social media as it was recently reported and some of the stages of intervention globally uh because this is one of the hot topics these days and there are four stages I'll discuss briefly with you and then uh where are those regulatory intervent mentions are in various countries and then I'll finish off with way forward.
I'm honored to to be invited and speaking at at the IPLC 2026 ABU intellectual property and legal committee conference this year with the team to explore AI copyright and royalties in digital content economy.
This is a timely discussion of safeguarding original content, tackling digital piracy, navigating navigating platformdriven distribution and protecting the future value of broadcast creativity. So my presentation falls under the sub team of navigating platformdriven distribution.
As you know most of our member APU member broadcasters and others uh now uh making use of social media and those social media platform to distribute their content. So uh so the while we are doing that our audience also get exposed to the social media content that are coming from other sources and that's why we we are very concerned about this.
First of all, I would like to thank Dr. Siman Tani Sharma, a legal manager at APU, who is a global expert in this field for inviting me to talk at your conference. I would also like to thank APU, which is not a foreign uh which is not foreign to me as I have been closely associated with APU with APU since 1997 in various roles. I'm glad to connect with you all today to address some of the key challenges that the broadcast and media industry faces today specifically in the in the area of social media social media ecosystem. So my presentation today focuses on protection of uh children in digital and social media ecosystem and regulatory responses.
I would like to emphasize five words from from that title the children and who get subject to these harms social media that's kind of a the cause for that and then how we protect them the protection and then regulation because some action is required for that so we have been discussing the harm uh that caused by cyerspace with the increased penetration of connectivity and devices because in the recent past connectivity um has increased. The affordability and the availability of devices have increased and this has brought the good as well as the bad. Though we have taken some action to mitigate such harm in the past over the last two decades. It can be observed that harms have increased in numbers and also impact they make on both children and families. Hence it is required to revisit this issue and challenges social media presents and take practical and effective measures.
My presentation mainly focuses on an action that globally looking at that banning of social media for children below 60. Yes, our previous discussion provided evidence of a wide range of harm that is caused to children by social media based on scientific studies by credible researchers and organization.
One of the worst harms of social media has led to the death of so many children. I want to remind you why I advocate for s for safe and responsible uh platform social media and digital engagements. I have communicated in my various articles and presentation that the technologies are advancing at a rapid pace such such that the the societies the governments and regulatory authorities cannot respond at the same pace.
Corporations, large tech companies and profit victimizing organization have no concern over the uh harm and the damage they cause to children 80% of society as large. I want to ensure new knowledge that's what we learn from pure research technology how we use that new technology to solve a problem and and and the engineering of it how we work on findings finding a sustainable efficient and effective solution to a problem are used to build a better lives and a better world I'm an advocate of sustainability and responsibility for betterment of people.
So then now the the harm due to the social media. So as you may aware already recently US courts ruled that harm caused to children and some of the people buy social media was due to the way social media was designed and not the content because uh the the current regulation that the US had given a kind of protection to the social media but uh because uh the the that regulation legislation says uh they they are not liable for the content. So it is not about the content this is about the design. So this is a very important junk in this one. So it was reported in a in a CNN online report that a California jury has found median YouTube liable on all the counts in a landmark case that accused the tech giants of intentionally addicting a young woman and injuring her mental health. And I quote, open quote, "Mita and YouTube were negligent in the design of their platforms, knew their design was dangerous, failed to warn of those risk and cause substantial harm to the plaintiff, the jury found."
Period. The the decision could set a precedence for hundreds of similar cases and lead to major changes in how social media platforms operate especially for young users as well as millions even billions in losses for the tech companies.
The case also marks watershed moment for social media in following years of concern from parents, advocates and lawmakers about online harms to children ranging from mental health and concern.
So um with this um I um carried out uh some research in the public domain to collect data uh we call data mining uh related to uh the banning of social media for under 16 years.
So uh when Dr. Samantha invited me to talk uh for your confidence on this topic. Uh that also prompted me to do uh the study and also uh it was uh further stimulated when uh when I heard in our media in in New Zealand we are asking young children whether they want access to social media or not. It is good of children. However, we should not co uh we should be cautious of weighing that input against what evidence-based research and educational sociologist proposed.
I learned recently expert treat this issue to a tobacco moment. I I'm sure you you know you know how the tobacco has caused harm and how many decades it took to act. uh it's not not not a do only the only the senior generation may remember how tobacco promoted decades ago especially in movies film society and how it is uh its consequences are still hurting families and societies.
So, so in this process of uh uh the uh the data mining I kept the the harm due to the social media. So as you as uh the these four stages uh globally uh one is the enforced the other one is passed and there are the legislation in progress. So that mean they are still drafting bills or discussing at select committees at parliaments like in New Zealand and there are other countries those who are proposing. So, so the uh when they are enforced, law is active uh and the the platform must comply and then when the legislations are passed and uh but the enforcement is yet to be and then the then the the progress is still still we are in the form of drafting uh P and uh to those regulatory interventions um for various countries. Um you could see the UK is the only country with powers granted on 9th of uh on the 9th of March this year and then there are four countries. It's good to note that we have Indonesia and Malaysia both countries from Asia Pacific the laws are in force. So among all four other countries and also the Australia which was the leading country who brought this idea up of banning social media for years 16 years and under and the bills are in progress in 10 countries number of European countries and also in US federal as well. So that means impacting all the states and also at the uh some other US states level as well.
New Zealand also we are in the select committee stage and then there are other countries that are proposed and I'm really looking forward to see how the European Union and also maybe EPU will on this exam progress. So with those um theme um some of my recommendation to to ABU and to you. ABU is the largest uh collective of broadcasters both public service media and commercial uh broadcasters commercial media and professional media institution. So we can collectively advocate and leverage our influence to address this vital issue to safeguard our children from social media harms. So I urge you to work with other collectives as well like EBU, European Broadcasting Union, African, Caribbean, North African as well and then all other.
So there are eight of those broadcasting unions under the umbrella of world broadcasting union. So we can work together to uh combat like we have done in many other issues. So let's work together and try to make some tangible advancement by our next uh IPLC conference in 2027. So with that uh I would say Takashi I thank you for patience and listening. Thank you once again and uh it was pleasure speaking to you and sharing my views. Uh so uh those are my personal views uh that I shared with you. Thank you. Uh >> thank you so much uh the doctor Amal uh thanks for your presentation and also advices uh they are really important to improve our organizations and our perspective. Maybe uh you mentioned technological developments are uh really rapid uh but regulation cannot follow. I guess it is one of the main problems for all of us because always uh regal legal regulations can follow but almost uh all the time in general we know the technological developments are rapid and we need to catch up. I guess all of us the main problems because all maybe broadcaster organizations always people ask what are we doing do we need to do some more things but uh also regulations uh problem. Thanks for all you all all our speakers and I guess we have some limited time to uh questions.
We are happy to uh hear some questions from our uh audience because our speakers I guess except uh doc Dr. Amma they are waiting us. Thank you so much.
Do you have any questions to ask our speakers?
M Ezra, can I just make uh a further recommendation?
Um last night uh there was a lecture in Sydney uh delivered by the editor and chief of Reuters, Alexandra Galoney. Um, and she talked quite a bit about the application of AI to news gathering and how there ought to be a rigorous licensing regime. Um, one of the interesting things that stood out for me was the the dimension of the costs we're talking about here that are no way are being remedied at the moment. She was saying that Reuters for example in Ukraine and Russia covering the conflict there at the moment they have 70 journalists working just in that territory on the ongoing story. Now the cost of that must be immense and what's happening is that um uh social media platforms are just scooping it all up and then uh AI operators um are wanting to use this material as well and also scoop it up. Um so this issue of licensing in the AI context is a is a very pressing one.
>> Thank you so much Mr. green. Uh I want to ask our audience uh do you have any questions or can I start maybe because I don't want to uh waited our speakers because maybe time can be really late for them. I'm not sure about this time differences.
Uh my first question is uh to uh Mr. green for you uh because uh you said that uh in your presentation in today's world uh many country many countries are arguing children protection regulation to adopt for example like my country Turkey as well uh but we know that there are many illegal tools like VPN many tools and children can uh learn about them very well maybe better than us unfortunately Absolutely. So, uh do you find these regulations can are successful or sincere? I would like to ask you Mr. Green. Thank you so much. Thank you.
The the central issue or one of the central issues is that of rigorous means of age identification.
And I mentioned how Malaysia is leading the way with its rigorous use of official documentation.
Um the difficulty that in some nations in some jurisdictions there is not the widespread use of official ID cards or ID identification.
I mean people will have passports but not uh or children will have passports.
um in nations where ID cards are part of the daily life that that that is a great step forward in terms of identification.
But in Australia the difficulty as I was outlining is that this is entirely based on self declaration. People can tell lies and say yes I'm under 16 but they uh they they they they may not reveal their real age.
Um, and parents have been known to take out accounts for their kids and there's no no penalty for parents acting in this way at present.
Thank you so much for your reply, Mr. Green.
Maybe I can ask another question to uh Mr. and Miss Arya from WIO because uh this topic is really interesting for me to be honest and uh you mentioned regulatory approach AI and copyright uh I'm really curious about your opinion what is the main uh problem uh to this approach uh to regulatory to make regulatory approach AI and copyright according your eyes what is the biggest problem. Maybe I would like to ask to you. Thank you.
>> So, Yash, do you want to respond or shall I?
Right, I'll go ahead.
Sorry, I think there's an echo.
But um so from the perspective of WIPO, one of the main challenges in terms of a regulatory approach on AI that we see is the contrast between the crossborder nature of content distribution ution and the national domestic nature of regulatory activities.
Certainly in the European Union that might be regional, there might be other regional approaches but in general there is a concern among both regulators and those who are being regulated about lack of consistency in the regulatory approach and the difficulties potentially in both implementing and complying with regulatory measures.
And linked to that is the question of transparency of knowing what content is being transmitted including across borders in order to once again apply regulatory measures.
Thank you so much miss completely agree because copyright is domestical so it's hard to implement and international aspect to balance is really hard especially at this point thank you so much again and uh I want to ask our audience uh I understand you are tired maybe you have any question or I can continue.
Okay, I I continue. And uh my another question to uh Mr. Evangelista uh in your presentation uh this sentences uh this uh protect this protection is interesting for me. You mentioned a right to rem remuneration for content for which no opt out was declared.
uh I'm curious about uh do you have any criteria or uh do you maybe study on something how can we reminerate them because we need to I how can I say determine some criteria uh to remineration I guess it is the maybe one of the uh biggest problem it will be because it can be sometimes especially legal aspect can be really problematic I would like to ask to you thank you so much >> yeah thank Thank you for your question.
But like we we base this uh proposal on the private copy levies uh solution that uh like we have in place in many countries all over the world. So kind of rely on system of collective management rights. So with the support of collecting societies >> I guess members uh can decide all of your their legislation or something.
>> Yes. Um well for instance we have uh our Nordic members they rely a lot on extended collecting licenses and they there is a great collaboration with the collecting societies for instance for the AI training we have in Denmark uh that new collecting society the DPCMO is responsible to collect the remaration when there is an AI uh training on content of the publishers.
and our our broadcaster member um operates under the umbrella of this uh uh cine society together with the publisher.
>> Thank you so much for your answer and thank you all of you. Uh these are really how can I say beneficial presentations for us. Thanks for your time and do you have any cons questions or >> MRA can I just make another just f brief final comment one of the discussion points in last night's lecture in Sydney uh was that the managing director of the ABC Hugh Marks mentioned that anthropic were in discussions with the ABC about licensing idiom Australian idioms ways of speaking speaking uh accents uh mannerisms uh particularly in terms of ABC drama content and things like that in order that that can be duplicated in an AI context. So I I found that really fascinating and takes the the whole issue of licensing just a bit step further.
>> Thank you Dr. Green. They are really valuable uh addiction uh for us. Thank you for your information and I guess no one has questions. Thanks for your time and I hope next time uh meet in person all of you.
>> Thank you. Thank you so much.
>> Bye-bye. Bye-bye.
>> Thank you.
>> Thank you so much. This is our last session today. Thanks for you all of you uh for your patience and your time. I guess I will give you >> Thank you so much. Uh thank you to all our speakers uh who joined us online. Uh it's always u very difficult for the secretary to find speakers who can come physically. uh but that's why we try to get all online speakers in the last session so that we can accommodate speakers who are with us here uh in the in the first three sessions. Uh so tomorrow we will start our meeting at 9:30 uh it will be on the sixth floor it won't be here it will be on the sixth floor uh and someone will be there to lead you inside the building just as today and I will now give the microphone to Harusan if he wants to make an announcement.
Thank you very much. And uh I'm sure that you are tired, so I will be brief anyway. And uh as I already announced, we're going to have supper this evening starting at 700 p.m. at a restaurant, namely Gloia uh in in Merur Saban. So uh uh I have already received you know uh your responses about that, but we I think we have spaces for table. So uh if you if you like to join jump in please feel free to uh feel please feel free to do that. Okay. So uh see you this evening or otherwise tomorrow morning.
almost.
Should we take a picture together?
Related Videos
54:37
Law
Session 5: Law, Genomic Medicine & Health Equity Conference/Webcast -- Dr. Consuelo Wilkins
UMNconsortium
234 views•2019-01-02
24:55
Law
Consider This: ICJ Climate Ruling — Holding Nations to Account
astroawani
307 views•2025-08-11
14:10
Law
The Third Gender I Legal Recognition of Intersex Community
KBCChannel1News
259 views•2023-07-09
3:52
Law
What Are Proceedings In Court? - CountyOffice.org
CountyOfficeLaw
129 views•2025-03-06
46:58
Law
Understanding the EU Electronic Evidence Package | Interview with Cristos Velasco | EP 7
DejanKosutic
816 views•2025-04-22
10:13
Law
8 Violations in ONE Day? Judge West Gives Ultimate Wake-Up Call
LawAndLogicTV
263 views•2026-04-26
2:22
Law
United States v. Navajo Nation Case Brief Summary | Law Case Explained
QuimbeeDotCom
589 views•2023-09-21
27:07
Law
Young Professionals: CPR Part 36 Offers with Harry Dyson
kingschambers4859
902 views•2024-04-30
Trending
9:05
Artificial Intelligence
2.4 BILLION Records Got Leaked...
DeepHumor
15K views•2026-07-22
13:25
Business
Playstation NO DISC/NO BUY Fight Is Over...
DavidJaffeGames
4K views•2026-07-23
19:32
Business
Should I buy a Sawmill?
essentialcraftsman
29K views•2026-07-22
16:52
Anthropology
Americans Confused in Australia for 17 Minutes Straight
IWrocker
17K views•2026-07-23