US Commerce Secretary Howard Lutnick has directed Anthropic to halt international access to its most advanced artificial intelligence models without explicit government permission.
The order, sent to CEO Dario Amodei, specifically targets the Fable 5 and Mythos 5 models. It mandates that Anthropic obtain a Commerce Department license before allowing any foreign national, anywhere in the world, to use the technology.
Lutnick grounded the directive in US export control laws covering civilian technology with potential military intelligence applications. The order, however, did not elaborate on the specific threat that prompted it.
Anthropic responded by disabling access to both models late Friday. The company has since held technical discussions with US officials to resolve the situation. The firm speculates the government acted upon discovering a possible jailbreak in Fable 5, a model already restricted from cybersecurity tasks. Anthropic criticized the order as disproportionate and warned such a standard could freeze new frontier model deployments industry-wide.
The unprecedented intervention follows Anthropic’s confidential IPO filing and a reported valuation exceeding $900 billion. The move introduces significant uncertainty for AI labs regarding international contracts and revenue. The license requirement remains in effect until further notice.