Anthropic released Claude Fable 5, the first public model from its powerful Mythos-class AI. Access to Mythos was previously restricted to select partners for cybersecurity research under Project Glasswing.
Fable 5 utilizes the same underlying architecture as Mythos but includes enhanced safety guardrails. These safeguards reroute high-risk queries regarding advanced cybersecurity and biology to the Claude Opus 4.8 model.
Anthropic conducted extensive internal and external testing to prevent jailbreaks of these safety features. The release aims to compete with frontier models while managing risks associated with highly capable AI systems.