U.S. Lifts Export Controls on Anthropic’s Claude Fable 5 and Mythos 5

US Commerce Department announced it had lifted export controls on Claude Fable 5 and Mythos 5 less than three weeks after suspending their use abroad. This restores global availability under revised national‑security assessments [Reuters].

Anthropic trained these models using constitutional artificial intelligence techniques, where each model critiques its own outputs against predefined principles rather than relying solely on direct human feedback. Experts say this self‑regulation can improve truthfulness and reduce harmful generation compared to conventional reinforcement learning from human feedback (RLHF) [EXPERT NOTES].

Earlier this year a security review flagged vulnerabilities related to adversarial prompt injection attacks that could manipulate system behavior predictability, especially concerning sensitive data handling protocols within enterprise environments. After addressing those risks, the department concluded the models meet prevailing regulatory standards applicable across jurisdictions worldwide [Reuters].

While lifting export controls broadens access outside U.S. borders, experts caution misuse remains possible. The decision underscores the need for robust oversight mechanisms in line with existing guidelines. Anthropic has previously emphasized caution surrounding deployment scenarios lacking strong safeguards.