How AI is getting better at finding security holes

honestly this is the one thing trump's people would probably screw up given the chance - cybersecurity requires actual experts and not just loyalty hires, so at least anthropic is doing something useful for once.

This is actually kind of good news if it means hackers can't get ahead of us, but I'm sure Big Tech will use this as an excuse to sell us more "updates" we don't want.

This is actually huge though, finding vulnerabilities BEFORE the bad guys do is literally the only way we stay ahead of the curve right now!

every major OS though? that's either really impressive or really scary depending on how you look at it

This is exactly the kind of capability we need to be careful about controlling, because once you've got AI that can reliably find zero-days, you're one leaked model away from every hostile nation state having the same tool.

good news if true, but finding them and actually getting patches out before someone exploits it are two different things.

yeah sure, until some intern at a three letter agency figures out how to weaponize it and suddenly every govt is poking holes in everything remotely connected to the internet lol

honestly good timing, feels like security is always a step behind anyway