Kinda makes sense.  Finding vulnerabilities seems like a task well suited for the AI, since they're not being asked to make any judgment calls, just scan through a massive amount of code and identify patterns that give rise to a vulnerability.

SimpleStacker

It is curious that most of the big AI companies rely so heavily on the apocalypse narrative. It's almost like they're trying to build a myth around their product.

Dunno.  This guy Nicholas Carlini is apparently a super legit security researcher and he's saying the models are much better than him at finding these sorts of vulnerabilities.  Seems legit.

news

Anthropic Claims New AI Model, Mythos, Is a Cybersecurity ‘Reckoning' - nytimes

Scoresby

> It is curious that most of the big AI companies rely so heavily on the apocalypse narrative. It's almost like they're trying to build a myth around their product.

Dunno.  This guy Nicholas Carlini is apparently a super legit security researcher and he's saying the models are much better than him at finding these sorts of vulnerabilities.  Seems legit.

https://www.youtube.com/watch?v=1sd26pWhfmg