Anthropic Releases Claude Fable 5, Its Most Powerful AI Yet, With Cyber Safeguards

This article discusses Anthropic's release of Claude Fable 5, the most capable AI model, featuring a split product strategy with safety classifiers.

• •Fable 5 (public) uses AI classifiers to detect and block cyber, biology, chemistry, and distillation requests, routing them to Opus 4.8, while Mythos 5 (restricted) maintains full capabilities for vetted defenders and costs $10/$50 per million tokens
• •Testing showed the model can autonomously identify and exploit zero-day vulnerabilities in major operating systems and browsers; external red teams found no universal jailbreak despite 1,000+ hours of testing
• •Fallback routing occurs in under 5% of sessions, making Fable 5 functionally equivalent to unrestricted Mythos 5 for legitimate users
• •Anthropic is implementing 30-day data retention to detect novel attacks and offering a Cyber Verification Program for vetted security professionals