security-measure

Anthropic added a new security guardrail to its Claude Fable 5 AI model, which helped convince the Trump administration to lift export controls; the safeguard redirects blocked requests to a less advanced model.