How Does Anthropic’s Mythos Claims Questioned by Cybersecurity Insider Work?

Anthropic’s Mythos is like a secret club that claims to know how a special robot thinks, but now someone who knows lots about keeping secrets online is saying maybe that club isn’t as sure as it sounds.

Imagine you have a robot friend named Mythos. This robot can talk, think, and even answer questions like a person. Anthropic says they know exactly how Mythos thinks, like knowing the exact recipe for your favorite cookie.

But now there's someone who works with secret codes and online safety (like a cybersecurity insider) saying, “Wait! Maybe we don’t actually know how Mythos thinks as well as we say.”

It’s like if you said, “I know the exact recipe for chocolate chip cookies,” but then someone who bakes all day says, “Hmm… I think there might be a few secrets in that recipe still.” They’re not saying it doesn’t work, just that maybe we don’t have the full story yet.

This person is asking questions like: Did they really test everything? or Is there something hidden? It’s like checking if your cookie jar is actually full, and finding out it might be a little empty.