Anthropic AI Blocks Security Researchers

San Francisco — Fable flags routine code reviews as cybersecurity threats. Anthropic's public Mythos model pauses any prompt it labels cybersecurity-related — including standard code review requests,

TechCrunch