Share Facebook Twitter LinkedIn Pinterest Email Anthropic reports the US government is aware of a technique that can bypass its Fable AI model, highlighting potential security vulnerabilities.