#mythos

Anthropic Built an AI So Dangerous It Refuses to Release It

Anthropic's Claude Mythos Preview found thousands of zero-day exploits in every major OS and browser. Why this AI cybersecurity model won't be released.