Anthropic launches Fable 5, a guardrailed version of its Mythos model designed for public use

Anthropic released Fable 5 on Tuesday with safeguards that prevent the model from answering questions about cybersecurity and biology, capabilities the company said made the unreleased Mythos model too dangerous for public release.

Fable 5 uses guardrails to restrict what the model is allowed to answer, though the underlying capabilities remain the same as Mythos; Anthropic tested the safeguards extensively with hackers and reported none were successful in bypassing them.

Early customers reported that Fable 5 significantly reduced the time it took to publish software and performed well on reasoning tasks (the step where an AI produces an answer).