The US Commerce Department on Friday cleared Anthropic to release its Mythos 5 frontier model to a tightly scoped list of roughly 100 companies and federal agencies, ending a two-week standoff that had left the company's two newest models offline. The clearance covers Mythos 5 specifically and follows direct negotiations between senior Anthropic staff and the Trump administration over the export-control directive issued earlier in June.
The freeze began with a June 12 letter from Commerce Secretary Howard Lutnick citing "national security authorities" and ordering Anthropic to suspend access to Mythos 5 and the lighter Fable 5 model. Anthropic complied within days, pulling both models from production and cutting off access "by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees."
“We have worked diligently to ensure America remains the global leader and AI while safeguarding our security”— Commerce Department spokesperson, US Commerce Department
That last clause is the part of the directive that shocked the industry. It applied not only to overseas customers but to Anthropic's own employees who hold foreign passports, an unusually broad reading of US export controls as applied to a software product. The company complied rather than litigate, then sent senior staff to Washington to negotiate a path back online.
Key facts
- 01Commerce Department cleared Anthropic to release Mythos 5 to roughly 100 companies and federal agencies on June 26, 2026.
- 02The clearance ends a two-week standoff that began with a June 12 letter from Commerce Secretary Howard Lutnick.
- 03Anthropic had disabled access to both Fable 5 and Mythos 5 earlier in June to comply with the export directive.
- 04The directive required suspension of access by any foreign national, including foreign national Anthropic employees.
- 05Senior Anthropic staff flew to Washington to negotiate the model's return; CEO Dario Amodei attended the G7 Summit on June 17 in Évian-les-Bains, France.
Friday's clearance is narrower than a full release. Mythos 5 goes to roughly 100 named US companies and federal agencies — not to Anthropic's broader API customer base, and not to international markets. Fable 5 was not mentioned in Friday's reports. The Commerce Department framed the decision in a statement to Reuters and Semafor as balancing US AI leadership with security concerns.
Lutnick, in a follow-up letter to Anthropic, said the company had "worked with the U.S. government to address risks associated with the Covered Models" since the June 12 directive. Neither side has disclosed what specific risk mitigations Anthropic agreed to, what evaluation results were shared, or what conditions attach to the 100-entity distribution list.
“Since the issuance of my June 12 letter, Anthropic has worked with the U.S. government to address risks associated with the Covered Models”— Howard Lutnick, US Commerce Secretary
The standoff played out against a backdrop of escalating US gating of frontier model launches. As AI Chat Daily reported earlier this month, OpenAI faced a parallel White House security review around GPT-5.6 that delayed and then narrowed the rollout. The Mythos 5 episode confirms the pattern: frontier model releases at both leading US labs are now subject to pre-deployment government review, with the deployment scope itself negotiable.
Anthropic CEO Dario Amodei spent part of the standoff outside the country. On June 17 he attended a working lunch with G7 leaders and global tech CEOs at the G7 Summit in Évian-les-Bains, France, where AI governance was on the agenda. The optics of a US lab CEO discussing AI policy with foreign heads of state while his own model was blocked by Washington were not lost on the people in the room.
Anthropic told CNBC during the freeze that both sides were "working quickly to get this resolved." The company has not yet publicly commented on Friday's clearance, and representatives did not respond to CNBC's request for comment. The full text of Lutnick's follow-up letter has not been released.
The narrow scope leaves real questions open. Anthropic's commercial business depends on broad API access; a 100-customer whitelist is closer to a defense-tech distribution model than to the company's normal go-to-market. Whether the company can negotiate a wider release — and on what timeline — will determine how much revenue Mythos 5 actually generates this year. Run-rate revenue at Anthropic has been a key part of the bull case for its near-$1T valuation discussed in recent reporting.
What this episode settles is that US frontier AI is now a regulated category in practice, even without a specific statute naming it as such. Commerce can freeze a model release on national security grounds, negotiate the scope of distribution, and approve a list of allowed customers — all on a two-week timeline, with no public rulemaking. For Anthropic and its peers, the operational implication is that release planning now includes a Washington review cycle as a hard dependency, not a courtesy. For the customers on Friday's list of 100, Mythos 5 just became one of the most exclusive pieces of software in the US enterprise.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
