US releases powerful Anthropic model Mythos to some US companies

10 hours ago

The U.S. government lifted the block on Anthropic's Claude Mythos 5 AI model, allowing its release to over 100 U.S. institutions.
This decision de-escalates a confrontation with the Trump Administration, which had imposed export controls on Mythos two weeks prior due to security concerns.
The status of the weaker Fable 5 model remains unclear, though talks suggest it may also be released later.
Commerce Secretary Howard Lutnick cited "appropriate safeguards" and Anthropic's commitment to collaborate with the government on protocols.
The move coincides with OpenAI releasing its GPT-5.6 model to government-approved partners.
The new arrangement removes licensing requirements for exporting Claude Mythos 5 to specified entities and their foreign employees.
This marks the start of a new regulatory regime giving the U.S. government control over frontier AI model releases.
The framework is being developed rapidly, with non-U.S. entities expressing frustration over dependence on Washington's decisions.
Initial U.S. concerns included Mythos being released to partners linked to China, such as a South Korean telecom provider.

Hasty Briefsbeta