Anthropic withholds Mythos Preview model because it's hacking is too powerful - Axios

This is the one of the first times an AI company has held back a model over societal risks.

You will be redirected in 10 seconds.