US Lifts Export Controls on Anthropic AI Models Mithos and Fable5

The US government has withdrawn export controls on Anthropic’s top artificial intelligence model, Claude Mithos.

Anthropic said on July 30 that the US Commerce Department had notified the company that it was lifting export-control guidance on Mithos5 and Fable5 after 18 days. The company said services for the two models would resume in July.

Commerce Secretary Howard Lutnick wrote in a social media post the same day that the government had worked closely with Anthropic for more than two weeks to analyze and approve Fable5. The move helped ensure consistency across the US government and strengthen US dominance in AI, he wrote.

Mithos5 drew attention from the security industry and governments worldwide for its ability to detect expert-level cybersecurity vulnerabilities. Fable5 is a version of Mithos5 with added safeguards that limit responses on sensitive topics that could be misused for hacking or weapons production.

The Commerce Department issued export-control guidance on June 12, saying the two models posed a national security threat and barring foreign nationals from accessing them.

Glenn Gerstell, a senior adviser at the Center for Strategic and International Studies, said export controls on innovative technology alone would not be enough to preserve an edge over China in advanced AI.

Hwang Jung-soo, Hankyung.com reporter hjs@hankyung.com