Anthropic’s Mythos Has Landed: Here’s What Comes Next for Cyber

Hinzugefügt: 2026-05-03

[00:00:05]Hello everybody and welcome to Reporters Notebook. I am Becky Bracken and I am here with my two colleagues to discuss uh this week's big blockbuster story, Mythos, the AI model to end all cybersecurity, and uh Glasswing, the forum that was established to sort of wrap industry and governments' head around it. I'm joined today by Eric Geller. Hello Eric. He is a senior reporter with Cybersecurity Dive. I'm sure you've read his byline on any number of big stories, as well as Phil Sweeney, who is a reporter with Cybersecurity at TechTarget Security.

[00:00:44]I'm sorry, we've rebranded. Is that correct?

[00:00:47]You got it. All right.

[00:00:49]Well, welcome both of you. I figured this was a pretty easy one for us to tackle. Do you, Phil, want to sort of walk us through the background as you understand it? For the Mythos preview, Anthropic developed it and and had some pretty startling success with it, things they did not expect. And before release, they'd said, "Okay, we can't do this. We can't release this. We need to talk about this and the applications for that." Especially security-wise, they found incredible volumes of zero days, unknown vulnerabilities, and some of them going back years. I think some they said many, in fact, are 10, 20 years old, not just a few outliers. There were um many, many that were going back many years undiscovered and uh their AI um model, the LLM, found them in almost no time at all. So, it was uh it was quite a jolt and as a result Anthropic has reached out to partners across the IT industry to try to come to some kind of consensus about, "Okay, what are we going to do about this before this becomes a major security crisis?" Eric, what do you What's the headline for you here?

[00:01:58]To me, this is a story about how the government is going to be increasingly dependent on the technology companies in a way that wasn't even really true in earlier phases of this kind of government-industry relationship. We think about cybersecurity as a domain where the private sector, because it runs the infrastructure, it has the best visibility, and the government is really dependent on it to understand cyberattacks. That's true. I think in the AI space, that reliance is even stronger because now it's not just that the AI companies have all this information about how hackers are trying to launch cyberattacks using their products, and you see Anthropic putting out that report last year about first AI-powered cyberattack. So, they have that visibility. They also have the ability, unlike say the critical infrastructure operators, to actually define the terms of the battlefield because it's their products that are being used to do some of this work.

[00:02:51]That's not to say AI is the only thing that hackers are using or the only thing that they need, but it is increasingly going to be part of the initial phase of an attack to use AI to figure out if your target has any vulnerabilities. And so, it's incumbent on the vendors to do as much as they can to prevent their tools from being weaponized in a way that really isn't true with a lot of other technology out there with the exception of like pen testing software, where we know that hackers use things like Mimikatz for for attacks. This is totally different. It's a totally different ballgame, and the government is entirely dependent on the vendors to not only make products that are not capable of being weaponized or that that aren't easily weaponized, but also to proactively, in advance, share with the government what they're finding and what their partners are finding with these tools. So, um you know, we're going to talk about Project Glasswing, and to me, what I'll be looking for there is as the companies use Mythos and discover vulnerabilities, what is the tempo of relationships with the government like?

[00:03:50]What's the tempo of information sharing with federal agencies like CISA? Is there something formally in place that says when a Glasswing partner finds a vulnerability, does it have to tell CISA? I don't think so. So, we're really seeing a an environment where these relationships haven't been well defined. How quickly that stuff gets ironed out is going to go a long way toward answering the question of how rocky are the next few years versus how smoothly does everybody kind of work together to prevent this weaponization of these tools?

[00:04:23]You are, Eric, the person that I look to to read the Washington DC tea leaves about uh what's going on in cyber.

[00:04:33]So, what is your uh sort of analysis of where we are? I mean, the the executive branch has been very clear that they want AI to run rampant and do nothing uh to hamper any kind of innovation.

[00:04:48]I'm hearing you say that might not be such a realistic outlook as we go forward for a national security.

[00:04:55]Am I hearing that right? And what is your How are those [snorts] things going to bump up against each other? How do you see this playing out in, let's say, the next 6 months? I think that's a pretty long runway in the AI. I mean, I do think that there's no real appetite in Washington to regulate what a company like Anthropic can do, in part because how would you define the boundaries of the regulation? How would you define safe behavior and unsafe behavior, safe coding and unsafe coding? I mean, if you define it based on the output, you know, can this can this tool help a hacker find a vulnerability, then you're going to be prohibiting a lot of behavior that we actually want to see because any tool that can help a hacker find a vulnerability can also help a defender find a vulnerability. There's no, you know, the technology is agnostic.

[00:05:40]There's no way to create an AI model that checks who you are, peers into your soul, and based on that decides whether it's going to tell you about a CVE in this in this internet-facing network appliance or what have you, right?

[00:05:52]That's That would be what we would want in a in a fantasy world, but that doesn't exist. So, you can't regulate the problem that exists. That's not to say you can't have any regulation. I'm not taking a stance here, but the idea that you can solve this particular problem through a regulatory framework, it's not like environmental pollution.

[00:06:10]You can't say only do the good things and don't do the bad things. That's not how the technology works. And I think you you know, I think you see policymakers recognizing that. And in the absence of kind of a regulatory answer, the the next best option is close conversations and collaboration so that as, say, Anthropic is finding out that its product can do something potentially dangerous, they're telling the government. And the government is deciding whether there's a thing that it can do to make the situation better or whether it ought to warn critical infrastructure operators, "Hey, there's this new tool coming out soon." That's That's not the sort of the the the perfect answer because again, the hackers can still use the tools, but it does at least speak to this idea of harm is going to happen. The best thing we can do to get ahead of it is to be talking to each other as we're learning about the harm. And that's not a satisfying answer, but I think it's kind of the best that Washington has at this point. A healthy answer, people talking is not happening just here and there, hither and thither right now. So, to see it happening here is important. But I also think it's worth touching on there is a real dearth of talent and expertise right now in government that I mean, there's just the experts that do exist are are working in these private sector businesses right now. Would you agree with that?

[00:07:27]Yeah, especially with all the layoffs that we've seen recently. And And I'm going to be looking to see what happens with NIST, the National Institute of Standards and Technology. They have an AI Safety Institute that was created in the last administration. It's been refocused now to really look at these core technical issues of the AI models and and the double-edged sword of this technology. So, I'm going to be very curious to see if that agency gets more involved in working side by side with the vendors to understand the implications of their product.

[00:07:54]And on the expertise front, Phil, enter Project Glasswing. And so, this is sort of this the the round table at which this conversation that Eric's been referring to is happening, correct? Tell us a little bit about what it is and what its parameters are.

[00:08:10]Right, right. It is a a group that is 12, excuse me, companies or organizations involved at the point of the spear. 40 or so others are going to be involved in in other ways. But yeah, the big ones you're talking about, you know, your cloud providers, AWS is involved here, Google is involved here, Microsoft.

[00:08:30]Anthropic itself, Apple is, Cisco is, CrowdStrike is uh uh JP Morgan Chase. It's a dozen big, powerful players in IT and finance and security and name it. So, they are getting access to the preview before any kind of release publicly. The idea being to give some sort of head start on fixing these vulnerabilities.

[00:08:54]It's uh it's an unusual level of cooperation. Rival companies will sometimes cooperate uh on some cyber security standards, interoperability, that sort of thing.

[00:09:05]There is, you know, the Linux Foundation, which is one of those foods, the Cloud Native Computing Foundation.

[00:09:11]They're They have kind of cooperative relationships across industries, but this this is uh there's a boldness here, an urgency that feels different um and is coordination on a scale that is rarely seen. So, especially among rivals, you know, some bitter bitter rivals in case uh you know, competitors.

[00:09:27]So, they're looking at um you know, um are this can't be fixed in Washington, it can't be fixed by individual companies. There has to be some sort of collective action. CrowdStrike CE CTO said something to the effect of, you know, this uh this needs to happen for defenders to unify, put these capabilities to work now before the adversaries can become involved in a serious way.

[00:09:50]And someone from Cisco supporting Glasswing said that, you know, the work is just too important to try try do and too urgent to do it alone. So, there is there's a sense here that this is a massive risk that's going to require massive effort to address. I wonder what you all make of this notion that this might be a bit overhyped. It's not lost on me that it's called mythos. It's not lost on me that a lot of this is very secret squirrel, but it's really big, but you can't see it. So, and I have seen some the AI Security Institute in the UK did a technical evaluation that found maybe it's not as potent of a the tool as they're making it out to be. A lot of the the criticism was a lot of the models that they ran it in weren't particularly well defended, really as well defended as even a mid-sized organization would be here. Nonetheless, it it represents a shifting tide. But, I wonder what you all make of this idea that maybe this is overhyped or that people are falling for what's essentially a marketing scheme cuz I have heard that. Well, I think partly it is true that the way you defend yourselves from the kinds of attacks that this tool can find is the same as the way you defend yourself from an attack that a human discovers and weaponizes and launches. Really, what we're talking about here is not the outcome. It's not the end result. It's not the kind of attack that gets launched, not for the most part anyway.

[00:11:19]It's the democratization of being able to do that work. If you're using default passwords, if you have um a network appliance that's out got out of date firmware, a human can exploit that if they know how to do so. And if they have if they spend the time to do so. This isn't something that only an AI can do.

[00:11:35]It's something that AI's making it easier to do. So, it's not as if AI has created new forms of attack. It's made it easier for more kinds of people with less knowledge to launch those attacks.

[00:11:46]You still need to be doing the same kinds of things you were doing in the past in terms of verifying your network perimeter, checking to make sure your user accounts are not being abused, identity as the perimeter, all these buzzwords that we know about and going to conferences over the years. These are still the things you need to do. And it just seems to me that people are focusing more on there's a tool that can help automate it rather than are the defensive techniques I need to implement the same as they used to be?

[00:12:13]And they are. They are the same. You need strong passwords just as you always have.

[00:12:17]Security hygiene, all the things.

[00:12:19]Absolutely. You need to do the same things you've always needed to do. It's just now you have to worry about more people trying to exploit your your failure to do those things.

[00:12:28]That's a great point. Phil, did you have anything to add to that? Just to add to that, yeah, there is I think a range of opinion and thought here. I think because this is somewhat unprecedented, you can't look at previous examples and say, "Oh, this is just like that." You know, so there's going to be some optimism, some skepticism, some cynicism even. I get that. But, what I would add is [clears throat] that um you know, if if we take Anthropic at its word, it has said that they had engineers with no formal security training just work with mythos preview and say, you know, find remote execution code vulnerabilities.

[00:13:01]And then boom, the next morning there would be a complete working exploit right there waiting for them.

[00:13:08]Um so, it does certainly lower the bar for sophistication. This can find and and also link together vulnerabilities and chain them in a way that um usually requires a lot of expertise, but um in this case, if what Anthropic is saying is possible, then that certainly changes and um and makes um cybercrime a a pretty um low bar for entry. Good point. Yeah, another smart point. One of the more delicious nuggets out of this that I've gotten excited about was the sort of timing of the release or the acknowledgement that the model could do some pretty amazing things along with a major hack of Chinese data swiped. And so, there was sort of maybe more twittering than, you know, hard reporting for sure that that maybe there the two were linked, you know, here tool of unprecedented, you know, danger falls in the laps of the American government and the next thing you know, the Chinese are getting their data swiped in in a huge big way. Do you think that are you hearing that there's any connection? What do you know about this if anything, Eric? I'd be very interested to learn more about that situation. I don't think at this point we have any reason to think it's connected to this tool. And in part because we think about about that Chinese organization as a target of nation-state espionage. If you think about the range of organizations that want to hack into that that entity, it includes the best hackers in the world. So, the idea that somebody could break into that organization, you don't need the advent of Claude mythos to explain that. If that had happened a year ago, two years ago, I would not have been surprised because the the people trying to get in are the best in the world. So, I do think the timing is just a matter of coincidence because you don't need Claude mythos to get in there if you are the the typical group trying to get in, which is NSA, CIA, British intelligence. Those are the groups trying to get in there and I I doubt that they are relying on Claude mythos to do their attacks.

[00:15:11]Okay, that is a more reasonable take than the delicious nature of the timing of the issue. That's good. And so, Phil, what are some of the questions that you are hoping your reporting will be able to answer about this moving forward?

[00:15:27]>> Right. Um I think it'll be interesting to see how the typical security organization that the the CISO and and and their teams, how they respond to this, how they react, how you know, if they're not among the special invitees for this um endeavor, um you know, what do they do to prepare? And um and so, I guess how all this branches out and and spreads throughout the security ecosystem will be interesting to watch how quickly that happens. There was something interesting that came out uh from the Cloud Security Alliance just the other day in response to all this that they wanted to give CISOs and and boards and executives otherwise to something to hang on to and say, "Okay, this is how you should be thinking about it even if you're not directly involved.

[00:16:07]This is going to change your life in a significant way, perhaps." They had uh some thoughts about, you know, sounding the alarm and being ready. And they said you know, "Prepare now, ask for more budget, hire more people, do more automation because there there will be a very shortened window between when a vulnerability is disclosed and the time when it can be exploited. You know, security teams are going to have to be ready to to step in and act quickly.

[00:16:35]I heard that at RSA quite a bit, attack at machine speed. And to me, that is the biggest question that teams are going to have to answer is the patching problem.

[00:16:45]They are going to have to get patching done at machine speed. And and I think there are still a lot of questions about how that is going to happen and it needs to happen yesterday. And I think practitioners are pretty well aware of that fact. It's just a matter of catching up to reality. Eric, I want to give you the final word here. What questions are you looking to get answered in your reporting?

[00:17:07]I'm very curious to see if this changes how the government thinks about its role in overseeing the sprawl of of AI technology. President Biden tried to get these companies to report to the government when they were doing red test red teaming tests and basically provide the results of those tests so that the government can in real time understand what's happening with the security audits that the companies are doing.

[00:17:29]President Trump got rid of that requirement. He described it as anti-innovation and too onerous and burdensome. But, I'll be interested to see if the Trump administration rethinks the hands-off approach it's taken to AI.

[00:17:40]I don't think it's going to completely rethink it, but I think there might be some folks advocating for a little bit more looking over the shoulder of some of these AI companies. Not with regulation, but with just some degree of oversight and input. Where would that come from? I mean, we're looking at an incredible shrinking CISO. You know, they're standing up a State Department sort of quasi cyber. There's NSA.

[00:18:04]There's I mean, where is this sort of thought leadership shift going to come from?

[00:18:12]Well, I don't think there's a lot of don't think there's a lot of appetite right now for it from anywhere, but there are some agencies that would be a natural fit to kind of have these kinds of interactions with the AI companies.

[00:18:21]NIST is the one that comes to mind because it's not regulatory. So, if you have the companies provide their reports to NIST, they're not worried that NIST is going to in in turn prosecute them, file a civil case, you know, they're not going to it's not like the FTC or the Justice Department where if you tell them about something, they might look at it and say, "You know what? You violated the law here. We're going to take you to court." That's not going to happen if you go to NIST because that's not the culture of the agency. So, I think it would be a good fit if they wanted to kind of bring something like the Biden executive order back into force. But again, I emphasize, I don't think there's a lot of appetite anywhere in the government for doing exactly what President Biden had in mind. Makes sense.

[00:18:59]Well, gentlemen, I have learned a lot today. Thank you so much for helping me understand this topic better and helping our audience understand it as well.

[00:19:08]Eric, where can we find more of your thoughtful deep reporting on this and other topics? Uh you can just go to cybersecuritydive.com.

[00:19:16]And Phil, tell us where we can find you.

[00:19:18]I'm at techtarget.com/security.

[00:19:21]My name is Becky Bracken. I am a senior editor with Dark Reading. You can find this along with every other sort of podcast and video and of course our deep thorough reporting at darkreading.com.

[00:19:35]Thank you all for listening. This has been another episode of Reporter's Notebook. We'll see you next time.