- •
Anthropic's Mythos model is too dangerous for release
“Notably, they are not releasing this model to the public because they claim it is too dangerous to do that. Instead, they are giving access to a consortium of tech companies, including Cisco, Broadcoms, or makers of Internet 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.”
- •
Mythos found a 27-year-old flaw in OpenBSD
“One of them was that this model apparently found a twenty seven 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 twenty seven years worth of professional security researchers had not been able to find.”
- •
AI can autonomously chain complex software exploits
“Alex Stamos said, like, yes. This is a big deal. And he was hoping for a long time that we would see a consortium come together like this because of exactly what you just said, Kevin. The intelligence in in these machines and their ability to work autonomously are now great enough that they can chain together exploits that human beings either would never see, would take them a long time to see, or they would just never get to because we're we're limited in ways that these machines are not.”
- •
Project Glasswing provides defensive access to tech giants
“You have a new model that you claim is the most powerful model in the world. So instead of selling it, you give a $100,000,000 of claud credits away to a consortium of companies that includes many of your competitors, which is what Anthropic is doing. That is not how I personally would market a spooky new model if I were in the business of marketing spooky new models.”
- •
Hard Fork Live returns to San Francisco June 10
“On June 10 in San Francisco, we are doing the second ever installment of Hard Fork Live. It's happening on June 10 in San Francisco at the Blue Shield of California Theater. Bigger venue than last year. Tickets will be on sale at nytimes.com/events. Not today, but next Friday, April 17.”
SECURE SOFTWARE
All podcast episode summaries matching SECURE SOFTWARE — aggregated across every podcast we track.
“Notably, they are not releasing this model to the public because they claim it is too dangerous to do that. Instead, they are giving access to a consortium of tech companies, including Cisco, Broadcoms, or makers of Internet 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.”
- •
Anthropic's Mythos model is too dangerous for release
“Notably, they are not releasing this model to the public because they claim it is too dangerous to do that. Instead, they are giving access to a consortium of tech companies, including Cisco, Broadcoms, or makers of Internet 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.”
- •
Mythos found a 27-year-old flaw in OpenBSD
“One of them was that this model apparently found a twenty seven 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 twenty seven years worth of professional security researchers had not been able to find.”
- •
AI can autonomously chain complex software exploits
“Alex Stamos said, like, yes. This is a big deal. And he was hoping for a long time that we would see a consortium come together like this because of exactly what you just said, Kevin. The intelligence in in these machines and their ability to work autonomously are now great enough that they can chain together exploits that human beings either would never see, would take them a long time to see, or they would just never get to because we're we're limited in ways that these machines are not.”
- •
Project Glasswing provides defensive access to tech giants
“You have a new model that you claim is the most powerful model in the world. So instead of selling it, you give a $100,000,000 of claud credits away to a consortium of companies that includes many of your competitors, which is what Anthropic is doing. That is not how I personally would market a spooky new model if I were in the business of marketing spooky new models.”
- •
Hard Fork Live returns to San Francisco June 10
“On June 10 in San Francisco, we are doing the second ever installment of Hard Fork Live. It's happening on June 10 in San Francisco at the Blue Shield of California Theater. Bigger venue than last year. Tickets will be on sale at nytimes.com/events. Not today, but next Friday, April 17.”
- •
Drift Protocol lost $285 million in exploit
“It's sitting at 250 million plus right now. Is that right, Tay? So Tay and I were talking about this before we started and I was like, it's not really a postmortem yet. It's like an active mortem or like something. So we don't know that much, I guess, about exactly what has happened here. So just probably leading with like, there's a lot of speculation, a lot of uncertainty, because this is like a couple of hours old.”
- •
DPRK hackers weaponize Zoom and Teams calls
“Oh, we, it's, it's, we've talked about it on the show. It's the Zoom calls. Well, now, now more often it's a Teams, it's a Microsoft Teams call, but it's exactly the same. So it's the exact same flow that we usually see in crypto. But they were targeting... developer, maintainer, yeah. That's the question is like, how the hell... it's very similar to crypto in the sense that you have individual people who are actually very smart, very talented, very capable, very computer knowledgeable.”
- •
Axios supply chain attack targets open source
“So basically what they did, they compromised the developer, they push a malicious version of the code to the Axios package, which is a dependency in like a bazillion other packages, like all of them. And so then anyone who installed any of these packages or updated any of these packages or ran, like updated or were working on any projects that had these packages in them, all those people were compromised. It's automatic, it's silent, and it's in the background.”
- •
Circle refuses to freeze stolen USDC funds
“The Drift Protocol hack was still unfolding when Kain, Taylor, and Luca went live. Within hours of a suspected admin key compromise, over $285 million had been drained across Solana, with Circle sitting on the ability to freeze the stolen USDC — and choosing not to.”
- •
Session token theft bypasses hardware MFA
“Even if you have hardware 2FA, protecting your authorization into your GitHub account or into your NPM account or whatever it is, when you authorize a token and save to your computer, that token is basically what you use to access these things for however long it is. If your computer is completely compromised in the way that DPRK compromises computers, that token, they take that token and they reuse it. Now it doesn't matter that you have MFA. It doesn't matter at all.”
Stay in the Loop
Free summaries of top podcasts. More signal, less noise.