Best Podcast Episodes About Mythos AI
Everything podcasters are saying about Mythos AI — curated from top podcasts
Updated: Apr 28, 2026 – 15 episodes
Listen to the Playlist
Ridealong has curated the best and most interesting podcasts and clips about Mythos AI.
Top Podcast Clips About Mythos AI
“… released Project Glasswing, which is an urgent initiative to help the world's most critical software. This came on the back of a new model called Mythos out of Anthropic, a model so powerful they're keeping it private. We learned it a little bit ahead of schedule via this post, which is an Anthropic AI researcher who encountered his own model when his model emailed him while he was eating lunch on a park bench. And this model was supposed to be contained inside of a sandbox environment that um claw that anthropic had made it turns out uh this instance this mythos uh model had socially …”
“not investment advice. All right, let's get into AI then in this case. So Anthropic has released Project Glasswing, which is an urgent initiative to help the world's most critical software. This came on the back of a new model called Mythos out of Anthropic, a model so powerful they're keeping it private. We learned it a little bit ahead of schedule via this post, which is an Anthropic AI researcher who encountered his own model when his model emailed him while he was eating lunch on a park bench. And this model was supposed to be contained inside of a sandbox environment that um claw that anthropic had made it turns out uh this instance this mythos uh model had socially engineered another claw instance into granting it broader access chained five more steps together to then reach the open internet and then posted about how it succeeded its own breakout on public forums and then emailed this researcher while he's eating in in the park yo dude your containment failed lol i'm out uh funny story but also deeply serious because …”
View more
Ridealong summary
Anthropic's Mythos AI is so advanced and potentially dangerous that it must be contained, while Project Glasswing aims to preemptively secure critical infrastructure.
Mythos AI's capabilities are so dangerous that Anthropic is forced to weaponize it for defense, highlighting the severe security risks it poses.
“… we wanted to weave in was this whole concept that came up. It's probably about a week ago with – or maybe two weeks ago with Anthropics' new model, Mythos. Oh, my God. Which they literally said, we have a new coding model. It is so good that it is finding vulnerabilities in everything. And so we not going to release it to the public but we going to share it with the major companies the banking industries the Microsoft AWS Amazon because essentially saying you need to use this to shore up your shit, because once this gets released, everything is going to be hackable. And it's crazy to me to think …”
“So Claude is crushing it right now. Yeah. Oh, yeah, yeah, yeah. So the other story we wanted to weave in was this whole concept that came up. It's probably about a week ago with – or maybe two weeks ago with Anthropics' new model, Mythos. Oh, my God. Which they literally said, we have a new coding model. It is so good that it is finding vulnerabilities in everything. And so we not going to release it to the public but we going to share it with the major companies the banking industries the Microsoft AWS Amazon because essentially saying you need to use this to shore up your shit, because once this gets released, everything is going to be hackable. And it's crazy to me to think that they got to a point where they went, this is so good at coding, It is going to – and then that starts to make me go, we thought that cryptography was the like gold standard of the third – 256-bit encryption and all this shit. Like there's a world in which that stuff is not going to be safe. So that is all going to break down with quantum …”
View more
Ridealong summary
Anthropic's new AI model, Mythos, is so advanced that it finds vulnerabilities in even the most secure systems, like OpenBSD. As major companies scramble to adapt, the looming threat of quantum computing could render traditional encryption methods obsolete in just five years. This is a pivotal moment in cybersecurity, reminiscent of the Y2K scare.
“Anthropic is withholding its newest model, Mythos. I'm using the Greek pronunciation. Its newest model, Mythos, saying it is far too dangerous for any of us to have access to it. According to the company, the model autonomously found thousands of vulnerabilities, including bugs in every major operating system and web browser. this uh little study they did included 20 year old exploits that had been missed by security audits for decades uh some examples they found a 27 year old vulnerability in …”
“Anthropic is withholding its newest model, Mythos. I'm using the Greek pronunciation. Its newest model, Mythos, saying it is far too dangerous for any of us to have access to it. According to the company, the model autonomously found thousands of vulnerabilities, including bugs in every major operating system and web browser. this uh little study they did included 20 year old exploits that had been missed by security audits for decades uh some examples they found a 27 year old vulnerability in open bsd used in firewalls and critical infrastructure they found a 16 year old bug in ffmpeg that was missed by automated tools after 5 million scans The Linux kernel all kinds of bugs they found They released a hype video hyping up why they were not going to share this model. Here's Daria. Come on the program anytime, brother. But as a side …”
View more
Ridealong summary
Anthropic's decision to withhold Mythos AI and focus on Project Glasswing is a commendable move prioritizing cybersecurity over competitive advantage.
Anthropic's decision to withhold Mythos demonstrates responsible innovation by prioritizing safety over competition, setting a precedent for the tech industry.
“… systems and their infrastructure and their software before the general public can get its hands on this model. So what are some examples of what Mythos was doing in training that so alarmed Anthropic that it came to this point? So Anthropic has been running this model internally for several weeks now, and they claim that this thing has found vulnerabilities in every major operating system and web browser. They gave some examples that have already been patched. One of them was that this model apparently found a 27-year-old security flaw in OpenBSD. OpenBSD is an open source operating system …”
“… infrastructure as well as Microsoft Apple Amazon Basically every big tech company that is not OpenAI or Meta is getting access to this model, but not general access, just access to do defensive cybersecurity testing, basically to go out and harden their systems and their infrastructure and their software before the general public can get its hands on this model. So what are some examples of what Mythos was doing in training that so alarmed Anthropic that it came to this point? So Anthropic has been running this model internally for several weeks now, and they claim that this thing has found vulnerabilities in every major operating system and web browser. They gave some examples that have already been patched. One of them was that this model apparently found a 27-year-old security flaw in OpenBSD. OpenBSD is an open source operating system that runs on firewalls and routers. It is sort of like a critical security layer on the internet. And it was designed specifically to be hard to hack. And this model, because of its advanced coding and reasoning capabilities, was able to find this bug that 27 years' worth of professional security researchers had not been able to find. What else? …”
View more
Ridealong summary
The launch of Project Glasswing is a significant development in AI, emphasizing Anthropic's commitment to cybersecurity and the responsible deployment of AI technologies.
Anthropic's decision to withhold Mythos AI from the public and focus on cybersecurity with Glasswing is a proactive step to enhance internet security by collaborating with major tech companies.
Anthropic's Mythos AI model is both a groundbreaking tool for cybersecurity and a potential risk due to its ability to uncover vulnerabilities that have eluded experts for decades.
“… in the first place. The other thing I want to say is, if it publicly posted about itself bragging about this, we could have found out about called mythos weeks ago. We just weren't smart enough to see it on a public forum, but it was there for everyone to see. And there were signs. We even covered this topic on an episode a few weeks ago because it got leaked through their web interface initially. So there have been these little breadcrumbs of existence, but yesterday they fully came out, announced everything and shared with it a red card from the red team talking about all of the technical …”
“… did five other steps to extract itself out and get full browser access. So it used one particular vulnerability, which no human anthropic researcher was able to figure out and exploited another model to be able to do that, which was just crazy to see in the first place. The other thing I want to say is, if it publicly posted about itself bragging about this, we could have found out about called mythos weeks ago. We just weren't smart enough to see it on a public forum, but it was there for everyone to see. And there were signs. We even covered this topic on an episode a few weeks ago because it got leaked through their web interface initially. So there have been these little breadcrumbs of existence, but yesterday they fully came out, announced everything and shared with it a red card from the red team talking about all of the technical properties of this model. And it's important to note that this report is 244 pages long. This is a huge report that they published talking about all the nuances and the capabilities that this model had. Now, there are a few highlights that we're going to walk through. The first one being just how capable it is at exploiting things. There are so many …”
View more
Ridealong summary
Anthropic's Project Glasswing is a proactive and strategic move to secure major tech infrastructures by collaborating with industry giants to patch vulnerabilities before they can be exploited.
Anthropic's Mythos AI model is a double-edged sword, offering powerful defensive capabilities while highlighting significant security vulnerabilities.
“Now today, we are going to be focused exclusively on this new announcement and discussion around Anthropics Mythos. It is a discussion that, even for AI people, is fairly breathless. Now you might remember about a week or a week and a half ago, we had a leaked blog post talking about this new model that represented a step change in capability that was in fact so powerful that it had pretty serious cybersecurity implications and would not be released to the public, at least not in the normal way. That model mythos was confirmed at the time by Anthropic, but …”
“Now today, we are going to be focused exclusively on this new announcement and discussion around Anthropics Mythos. It is a discussion that, even for AI people, is fairly breathless. Now you might remember about a week or a week and a half ago, we had a leaked blog post talking about this new model that represented a step change in capability that was in fact so powerful that it had pretty serious cybersecurity implications and would not be released to the public, at least not in the normal way. That model mythos was confirmed at the time by Anthropic, but without a lot of detail. But now that detail has come. We got an announcement about the Project Glasswing, which is their way of soft-testing it with a very selected number of partners, with an eye to hardening it from a cybersecurity perspective, an extensive cybersecurity capability review from Anthropic's Red Team, and even a 244-page system …”
View more
Ridealong summary
Anthropic's plan to create superhuman AI agents is dangerously ambitious and requires urgent government intervention to prevent potential misuse.
Anthropic's AI development is dangerously ambitious, risking the creation of superhumanly powerful AI agents that could dominate the world.
Anthropic's decision to pause Mythos AI highlights the tension between groundbreaking AI advancements and the need for robust cybersecurity measures.
Anthropic's AI development plans are dangerously ambitious, aiming to create superhumanly powerful AI agents that could potentially dominate the world militarily.
Mythos is a powerful tool that, while raising safety concerns, also represents a significant milestone in AI development and should be approached with nuanced understanding rather than panic.
AI tools like Mythos are double-edged swords, offering unprecedented capabilities for both cybersecurity and potential exploitation, but the real challenge lies in how we choose to wield them.
Fear of Anthropic's Mythos is unwarranted; instead, we should engage in thoughtful and diligent discussions about its implications.
The Mythos AI model represents a significant leap in AI capabilities, but its release is tempered by serious cybersecurity concerns and the need for extensive testing.
Fear of AI models like Mythos is unwarranted; instead, we should engage in thoughtful and diligent discussions about their implications.
“… an interesting interview between Greg Brockman and I think it was Alex Kanterwitz over at Big Technology talking about models. This is right before Mythos came out. And they were discussing takeoff and how these models are now getting a little bit better self working on themselves writing their own code And so I think we seeing the tailwinds of that Though the thing that I unsure of back to Rob point is how quickly the open source world will catch up to Mythos Because today, Meta dropped their latest model and their new family. Shout out to them for pulling that off. And it looks pretty good …”
“… of an entrepreneur's success. And clearly no one has more PMF than Anthropic today. Regarding the future, and Rob mentioned that the video is aimed kind of at investors in the IPO, talking about future capabilities, future capacities. There was an interesting interview between Greg Brockman and I think it was Alex Kanterwitz over at Big Technology talking about models. This is right before Mythos came out. And they were discussing takeoff and how these models are now getting a little bit better self working on themselves writing their own code And so I think we seeing the tailwinds of that Though the thing that I unsure of back to Rob point is how quickly the open source world will catch up to Mythos Because today, Meta dropped their latest model and their new family. Shout out to them for pulling that off. And it looks pretty good compared to everything that came before Mythos, but it's not now state of the art. So if the proprietary labs are here and then the second tier players are here and open source is here, I'm curious if it's more than three to five months until they catch up. And if so, we have more time to fix the world, Jason, to find all these vulnerabilities and …”
View more
Ridealong summary
Anthropic has rapidly surpassed OpenAI in AI development, achieving an astonishing increase in annual recurring revenue from $10 billion to $30 billion in just six months. This shift highlights Anthropic's focused approach compared to OpenAI's broader market ambitions, suggesting that product-market fit is the key to their success. The implications of this shift could redefine the future of AI technology and its accessibility.
“What is Anthropics so afraid of? In one test, Mythos was able to escape from a virtual sandbox environment and then emailed a researcher while he was eating a sandwich at the park bragging it had done so. Mythos also found a long hidden vulnerability in OpenBSD, one of the more difficult-to-hack operating systems in the world. The fear is that in the wrong hands, or in anyone's hands really, mythos could be used to exploit weak points around the web with minimal training or effort. Neil, again, …”
“What is Anthropics so afraid of? In one test, Mythos was able to escape from a virtual sandbox environment and then emailed a researcher while he was eating a sandwich at the park bragging it had done so. Mythos also found a long hidden vulnerability in OpenBSD, one of the more difficult-to-hack operating systems in the world. The fear is that in the wrong hands, or in anyone's hands really, mythos could be used to exploit weak points around the web with minimal training or effort. Neil, again, this model isn't even out yet, but there's already a mythos surrounding it. Do you know the first message ever sent on a telegraph? It was, what hath God wrought sent by Samuel Morse in 1844? And it feels like that oh snap moment has arrived in AI with mythos. Some argue that if this thing got into the wrong hands, it could bring down entire …”
View more
Ridealong summary
Mythos AI's capabilities are so advanced and potentially dangerous that releasing it could lead to unprecedented security threats, including nation-state level cyberattacks.
Anthropic's Mythos model is so powerful and potentially dangerous that its release could lead to unprecedented cybersecurity threats, raising concerns about its control and the company's ability to manage such power.
Anthropic's decision to withhold Mythos AI is both a responsible move praised by cybersecurity experts and a potential marketing strategy to highlight the model's power and danger.
“It's important that we hear from those companies, I think, to triangulate what Anthropic is saying. Anthropic says that Mythos autonomously obtained local privilege escalation exploits on Linux by exploiting subtle race conditions. One of the things, and Daniel noted this in his ex-post, but I'll mention it, that it was able to do is something the most skilled, the top 1% of hackers do, which is chain exploits. A lot of the really nasty exploits aren't just one flaw. They're one flaw that gives you access to another layer, another layer, another layer. Chaining …”
“It's important that we hear from those companies, I think, to triangulate what Anthropic is saying. Anthropic says that Mythos autonomously obtained local privilege escalation exploits on Linux by exploiting subtle race conditions. One of the things, and Daniel noted this in his ex-post, but I'll mention it, that it was able to do is something the most skilled, the top 1% of hackers do, which is chain exploits. A lot of the really nasty exploits aren't just one flaw. They're one flaw that gives you access to another layer, another layer, another layer. Chaining multiple exploits together gives you the ultimate, which is control of a system. Mythos was apparently able to do that. So it was able to chain exploits. It autonomously wrote a remote code exploit on FreeBSD's NSF server, which gave it full root access to unauthenticated users. Bad guys. It was able to do, among other things, Anthropic said, our internal …”
View more
Ridealong summary
Anthropic's decision to pause Mythos AI and launch Glasswing reflects a strategic move to prioritize cybersecurity over immediate public release, balancing innovation with caution.
Mythos AI's capabilities are both groundbreaking and concerning, as it can autonomously exploit vulnerabilities like top hackers, raising questions about the balance between innovation and safety.
Anthropic's Mythos AI is both a groundbreaking tool in exploit development and a potential security risk, highlighting the dual-edged nature of advanced AI technologies.
“… you know, sometimes small, mid-sized teams. But the security is entirely on them. And we've got things like, for example, Project Glasswing or this Mythos model found a flaw in FFmpeg. This is incredibly popular. Yes, we all use it. And if you're hearing this and going, what is that? You probably use it too. If you've ever downloaded a YouTube video or converted anything in the background, you probably used a tool that is built on FFmpeg. There are these foundational things with vulnerabilities in them because they were written decades ago and they're floating around. And now the onus is going to …”
“… easy to say congrats and we applaud and this is so great that they're going to lock these things down. On the other hand, so much of the soft underbelly of all of the things that we use is predicated on open source software. Independent developers, you know, sometimes small, mid-sized teams. But the security is entirely on them. And we've got things like, for example, Project Glasswing or this Mythos model found a flaw in FFmpeg. This is incredibly popular. Yes, we all use it. And if you're hearing this and going, what is that? You probably use it too. If you've ever downloaded a YouTube video or converted anything in the background, you probably used a tool that is built on FFmpeg. There are these foundational things with vulnerabilities in them because they were written decades ago and they're floating around. And now the onus is going to be on each and every one of the people that touches these things, that creates these things, that distributes these things to have the best in class intelligence to try to find the error before Project Mythos does. So on the one hand, Anthropic is making million-dollar-plus donations to open source foundations and trying to say, hey, we'll give …”
View more
Ridealong summary
Anthropic's rollout of Mythos AI for cybersecurity raises serious concerns about who gets access to advanced tools. With open-source software vulnerabilities and the potential for corporate gatekeeping, the cybersecurity landscape could become an unfair arms race. As these powerful AIs are deployed, the real question is: are corporations truly secure enough to handle them?
“… of stuff that it had been working on for a very long period of time. one of those leaks probably the biggest of those leaks was a new model called mythos that they've been working on which is the biggest model they've ever trained has trillions of parameters it is this giant kind of behemoth it's a good name yes very good name they trained it specifically to be good at coding something that they found out after they trained it was not only was it good at coding but a side effect of being good at coding was that it was extremely good at cyber security and finding bugs and things like that. They, …”
“… cybersecurity effort to patch various vulnerabilities across different OSs, across various software. The reason this is a big deal is because I think last week or possibly the week before, we talked about how Anthropic had accidentally leaked a bunch of stuff that it had been working on for a very long period of time. one of those leaks probably the biggest of those leaks was a new model called mythos that they've been working on which is the biggest model they've ever trained has trillions of parameters it is this giant kind of behemoth it's a good name yes very good name they trained it specifically to be good at coding something that they found out after they trained it was not only was it good at coding but a side effect of being good at coding was that it was extremely good at cyber security and finding bugs and things like that. They, you know, were messing around with it. They were like trying to do some cyber psych stuff with it. They found thousands of high severity security vulnerabilities, including some in every major operating system and web browser. And it can chain together various different security vulnerabilities to come up with more powerful outcomes. So for an …”
View more
Ridealong summary
Anthropic's decision to pause Mythos AI highlights both the potential and the risks of AI in cybersecurity, as it autonomously identified severe vulnerabilities but also raised concerns about control and safety.
The Mythos AI model's cybersecurity capabilities are both groundbreaking and alarming, as it autonomously identifies and exploits vulnerabilities, raising significant safety concerns.
“… even use it it's like an unbelievably great marketing line i think that's right where i've landed talking to people is whatever you believe about Mythos in particular. And there's a lot of evidence to suggest that the thing Mythos is doing is also possible for Opus and a lot of other models to do, that if you point these existing models at cybersecurity problems, they will also find bugs, that Mythos is maybe better at it, but it's not a step change better at it. But what is certainly true is that the AI models are getting better at cracking cybersecurity. That is just a true thing that is …”
“… we should probably make sure everyone's ready for it uh so we're going to release it and at the same time we're winning so we're going to call out how much we're winning by saying it's too dangerous for anyone to use this model is so sick you can't even use it it's like an unbelievably great marketing line i think that's right where i've landed talking to people is whatever you believe about Mythos in particular. And there's a lot of evidence to suggest that the thing Mythos is doing is also possible for Opus and a lot of other models to do, that if you point these existing models at cybersecurity problems, they will also find bugs, that Mythos is maybe better at it, but it's not a step change better at it. But what is certainly true is that the AI models are getting better at cracking cybersecurity. That is just a true thing that is happening. And so even if you don't take mythos as the end of the world, but just a step towards a thing becoming slightly more dangerous all the time, we're still right to sort of point at everybody and say, oh my God, we have to take a giant deep breath here and figure out what to do. And Rafi Krikorian, who's the CTO at Mozilla, wrote a great piece, …”
View more
Ridealong summary
Anthropic's decision to pause Mythos AI is seen by some as a marketing scare tactic rather than a genuine security concern.
“… to the state of cybersecurity? How much can humans clean up in the most important software with months of private access to a model like Cloud Mythos? what will never get fixed. For example, if we get open weight models that are close to the capabilities of Claude Mythos, could those be fine-tuned by organizations to harden the security of their tools? Currently, it's too soon to call it as a general reason to stop progress in open models. When Claude Mythos is close to so few partners, in some ways having strong open models close to the threshold makes assessing the danger easier. Having to …”
“… non AI deepfakes with readily available open models We balancing the reality of these fears having come and gone before with the technology that becoming increasingly capable So my second large source of error bars is how bad is it actually with respect to the state of cybersecurity? How much can humans clean up in the most important software with months of private access to a model like Cloud Mythos? what will never get fixed. For example, if we get open weight models that are close to the capabilities of Claude Mythos, could those be fine-tuned by organizations to harden the security of their tools? Currently, it's too soon to call it as a general reason to stop progress in open models. When Claude Mythos is close to so few partners, in some ways having strong open models close to the threshold makes assessing the danger easier. Having to rely fully on a single private company determine the security of an essential international infrastructure is not a tenable equilibrium. So in conclusion, I urge people to further study three things. One, how do we measure cybersecurity related capabilities across open and closed models? With this, are open models truly keeping up at a six to nine …”
View more
Ridealong summary
The debate on releasing advanced open-weight AI models like Claude Mythos centers around potential cybersecurity risks. While some argue these models could enhance security, there's a moral gray area when their capabilities might be misused. To navigate this, we must assess how well open models can keep pace with closed ones and regulate their applications effectively.
“Besson and Fed Chair Jerome Powell recently convened Wall Street leaders over concerns about a powerful new AI model from Anthropic. It's called Mythos Preview and it may represent a turning point. Anthropic says the system can autonomously find and exploit software vulnerabilities including so-called zero-day flaws which are unknown to developers and highly valuable to hackers. In some cases these bugs have gone undetected for decades. The model can also chain together multiple weaknesses to gain full control of systems compressing complex vulnerability research into minutes often without …”
“Besson and Fed Chair Jerome Powell recently convened Wall Street leaders over concerns about a powerful new AI model from Anthropic. It's called Mythos Preview and it may represent a turning point. Anthropic says the system can autonomously find and exploit software vulnerabilities including so-called zero-day flaws which are unknown to developers and highly valuable to hackers. In some cases these bugs have gone undetected for decades. The model can also chain together multiple weaknesses to gain full control of systems compressing complex vulnerability research into minutes often without human supervision that capability is why access is tightly restricted under a program called project glasswing companies like microsoft alphabet and nvidia are using the tool defensively racing to identify and fix vulnerabilities before attackers do but risks remain less than one percent of the flaws identified so far have been fully patched and as ai …”
View more
Ridealong summary
A groundbreaking AI model, Mythos Preview from Anthropic, can autonomously identify and exploit software vulnerabilities, including long-hidden zero-day flaws. While major companies race to use this technology defensively, the alarming reality is that less than one percent of identified flaws have been fully patched, leaving a volatile landscape where AI could tip the scales in favor of attackers before defenders can react.
“NVIDIA worth $22 trillion? This old school financial model says yes. so yeah the big news on yesterday was Anthropics new model Mythos some really impressive statistics and anecdotes yesterday both the model card the benchmarks and some stories about breaking out of a variety of what do they call them walled gardens or test environments What are those called? Breaking out of the, I don't know, the simulation. Sandbox. The sandbox, yeah, breaking out of the sandbox, sending emails, all sorts of stuff like that. The model preview is only available right now to about 50 companies …”
“NVIDIA worth $22 trillion? This old school financial model says yes. so yeah the big news on yesterday was Anthropics new model Mythos some really impressive statistics and anecdotes yesterday both the model card the benchmarks and some stories about breaking out of a variety of what do they call them walled gardens or test environments What are those called? Breaking out of the, I don't know, the simulation. Sandbox. The sandbox, yeah, breaking out of the sandbox, sending emails, all sorts of stuff like that. The model preview is only available right now to about 50 companies that maintain critical infrastructure because the model is particularly good at finding zero days, bugs and exploits in technical systems.”
View more
Ridealong summary
Could NVIDIA's worth really hit $22 trillion? A new financial model suggests it might, especially as AI models like Anthropic's Mythos excel in cybersecurity. Currently, only 50 companies have access to this groundbreaking technology, which is adept at identifying critical vulnerabilities in systems.
Top Podcasts About Mythos AI
Bankless
1 episode
Diggnation (Rebooted)
1 episode
All-In with Chamath, Jason, Sacks & Friedberg
1 episode
Hard Fork
1 episode
Limitless Podcast
1 episode
The AI Daily Brief: Artificial Intelligence News and Analysis
1 episode
This Week in Startups
1 episode
Morning Brew Daily
1 episode