U.S. allows Anthropic to release Mythos AI to ‘trusted’ US organizations
SkimNews Take
The "trusted institutions" carve-out creates a tiered frontier AI ecosystem, where the most capable models remain accessible only through government-cleared channels rather than the open market.
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- U.S. government lifted its two-week block on Anthropic's Claude Mythos 5 model, with Commerce Secretary Howard Lutnick notifying Anthropic's chief compute officer Tom Brown in a Friday letter that the model can now reach more than 100 US institutions, including major companies and government agencies
- Fable 5 remains unreleased — the Lutnick letter is silent on the weaker companion model — though people close to the talks said they are moving toward releasing it on an unclear timeline
- The original block was triggered by warnings from Amazon and other companies that Mythos could be "jailbroken" for malicious purposes; Lutnick cited "significant progress" in daily government-Anthropic talks and Anthropic's commitment to work on protocols and release standards
- A license will no longer be required to export, reexport, or transfer Mythos 5 to entities listed in Annex A of the letter, including their foreign national employees
- OpenAI released its latest model, GPT-5.6, to a short list of government-approved partners the same day
- The Commerce Department framed the move as the beginnings of a new regulatory regime giving the US control over frontier AI model releases; European officials and other US allies have expressed frustration at their new dependence on decisions in Washington
- The US had raised concerns that Mythos was released to partners too closely linked to China, reportedly referring to a South Korean telecommunications provider
Why it matters: Lutnick's letter creates a new template: Washington can block a frontier AI model for two weeks, negotiate safeguards, and then greenlight release only to a curated US institution list. For Anthropic, the immediate gain is resumed access for 100+ partners, but every future frontier release now runs through Commerce. For non-US governments and companies already expressing frustration, the de facto gatekeeper for cutting-edge American AI just became the US government itself.
Ask SkimNews
