Anthropic allowed to release Mythos AI to some companies, agencies

The U.S. authorities on Friday granted Anthropic permission to launch its Mythos 5 mannequin to a gaggle of roughly 100 corporations and federal companies, CNBC has confirmed.

The choice, detailed by the Commerce Division in a letter to Anthropic, marks a significant step ahead within the negotiations between the Trump administration and the artificial intelligence firm, which have been engaged in a two-week standoff over its newest fashions: Fable 5 and Mythos 5.

“I’ve decided that acceptable safeguards are in place to allow sure trusted companions to entry the Claude Mythos 5 Mannequin,” U.S. Commerce Secretary Howard Lutnick wrote within the letter, which was considered by CNBC.

The letter, addressed to Anthropic co-founder Tom Brown, didn’t grant Anthropic approval to revive entry to Fable 5. Brown has reportedly taken the lead in negotiations with the Trump administration, changing Anthropic CEO Dario Amodei in that function.

Anthropic mentioned earlier this month that it disabled entry to its Fable 5 and Mythos 5 fashions to adjust to an export management directive from the federal government that cited “national security authorities.” The corporate mentioned it was instructed to droop all entry “by any overseas nationwide, whether or not inside or exterior the USA, together with overseas nationwide Anthropic workers.”

The surprising transfer landed simply days after Anthropic announced the fashions, which the corporate touted as state-of-the-art throughout numerous completely different business benchmarks. Fable 5, particularly, marked the primary time that Anthropic launched such a complicated providing to the general public, because of new safeguards that block responses in particular high-risk areas.

Earlier on Friday, Anthropic rival OpenAI introduced three new AI fashions and mentioned it is complying with the U.S. authorities’s request to initially restrict the rollout to a “small group of trusted companions.”

OpenAI mentioned in a weblog publish that it “believes in broad entry” and is working to make the fashions — GPT-5.6 Sol, Terra and Luna — typically out there within the coming weeks. OpenAI mentioned it previewed the fashions’ capabilities and shared its plans with the federal government forward of Friday’s launch.

Of the 2 corporations, Anthropic has had a more durable time coping with the Trump administration. Earlier this 12 months, the Division of Protection clashed with the corporate over disagreements concerning use of its fashions.

After negotiations between the 2 sides collapsed, the DOD declared Anthropic a provide chain danger, which means the corporate purportedly threatens U.S. nationwide safety. The label has traditionally been reserved for overseas adversaries, and requires protection contractors to certify that they won’t use Anthropic’s Claude fashions of their work with the army.

Anthropic sued the Trump administration in an effort to reverse its blacklisting, and litigation continues to be ongoing.

WATCH: Cyber experts stand behind Anthropic models in White House face-off