Claude Mythos

Anthropic keeps new AI model private after it finds thousands of external vulnerabilities

Anthropic’s most capable AI model has already found thousands of AI cybersecurity vulnerabilities across every...