Anthropic Releases Claude Fable 5 and Mythos 5 Models

Original: Anthropic Offers Mythos Upgrade for Cyber Partners and a ‘Safe’ Version for the Rest of You

Why This Matters

Demonstrates how AI companies balance advanced capabilities with safety concerns in cybersecurity applications

Anthropic released two AI models Tuesday: Claude Fable 5 for public use with cybersecurity guardrails, and Claude Mythos 5 for select partners. Fable 5 blocks security-related queries, routing them to older Claude Opus 4.8 model instead.

Anthropic launched Claude Fable 5 publicly and Claude Mythos 5 to limited partners Tuesday, upgrades from April's Mythos Preview model. Both use the same underlying model, but Fable 5 includes guardrails blocking cybersecurity, biology, and chemistry questions, redirecting them to Claude Opus 4.8. Mythos 5 remains restricted due to vulnerability-discovery capabilities that could aid cyberattacks. Head of product Diane Penn said the approach emerged as 'most viable' after testing since April. The company is collaborating with the US government on Mythos rollout and provides unrestricted access to Project Glasswing partners and select biology researchers. Anthropic plans a future trusted access program and acknowledges competitors will eventually offer similar capabilities.

Source

wired.com — Read original →