I don't feel like there's nearly enough evidence here to know whether the claims are accurate. The group claimed they accessed it and used screenshots as evidence, whereas Anthropic says that they are looking into it but haven't found any evidence of it so far. I don't feel like screenshots are particularly strong evidence, but I also wouldn't expect Anthropic to admit to it if they did realize it happened either.
by motbus3
0 subcomment
As other mentioned, there is zero evidence of that, but of the company who claims did something so smart cannot use it to protect itself, it seems it is worthless.
Unless this is all lies to make people think Skynet is on the loose. And by now, anthropic and other aislopware are full of lies
by overflowy
1 subcomments
At this point, this all feels like marketing, people getting paid left and right to make all kinds of claims about this model as long as it stays in the spotlight.
by drewfax
0 subcomment
This feels like a PR stunt by Anthropic. We understand, Mythos is a great model. But please stop unneeded publicity for the sake of IPO.
Hacker News