Jack Crivici-Kramer

In Anthropic right now, the company, they're calling this the sandwich incident.

Yeah, because the AI was like, hey, I'm out.

Hope you're having a good lunch.

Yeah, the guy was in the middle of a BLT and discovered that he had a world-changing AI get out on the loose.

And then the AI actually bragged about it escaping by posting on a publicly visible blog that he got out.

So clever, in fact, that Anthropic made an unprecedented decision.

Yes.

They won't launch this AI model to the public.

Because it's just too intelligent.

It's too intelligent.

So on Tuesday, Claude published a 255-page report about how this mythos AI is too smart for its own good.

The TLDR is that it's dangerously smart.

Ugh.

Like Stuyvesant High School smart.

Yeah, yeah.

And if it gets into the wrong hands, that is a huge cybersecurity risk.

And here's the proof which Anthropic published.

Yep.

This model has already found thousands of cybersecurity vulnerabilities in virtually every major operating system on Earth.

This AI brought the receipts.