US allows Anthropic to release Mythos AI to 'trusted' US organizations

WASHINGTON: Anthropic said on Friday that the U.S. government has allowed it to release ​its powerful Claude Mythos 5 artificial intelligence model to some “trusted” U.S. organizations, partially reversing a two-week-old order to suspend access over national security risks.

More than 100 ‌companies and institutions will now have access to Mythos 5, including many Fortune 500 companies, a source familiar with the new directive said, declining to be identified due to the sensitivity of the matter.

Concern that powerful AI systems could be misused by military intelligence users in China, Russia, or other countries of concern has prompted President Donald Trump’s administration to take an aggressive approach to oversight of releases of Anthropic’s and rival OpenAI’s frontier models.

OpenAI said earlier in the day that it ​was delaying a full public launch of GPT-5.6 at the U.S. government’s request, limiting its access to a small group of vetted partners whose details were shared with the authorities.

Anthropic ​had abruptly disabled its most advanced AI models — Mythos 5 and Fable 5 — for all users after the government’s June 12 export control order.

“Today, the government notified ⁠us that Mythos 5, our strongest cybersecurity model, can be redeployed to a set of US organizations that operate and defend critical infrastructure,” Anthropic said in a statement on Friday.

“We’re restoring access for these ​organizations quickly, and we’re continuing to work with the government to expand access to Mythos 5 and make Fable 5 available for general use again,” it added.

GOVERNMENT CRITICISED FOR PICKING WHO GETS ACCESS

The government’s vetting ​of which companies can gain access to Mythos has drawn much criticism.

“No one knows how these companies are picked and why everyone else is excluded,” said John Coleman, legislative counsel for the Philadelphia-based nonpartisan free speech organization, the Foundation for Individual Rights and Expression.

“This is putting too much power in the hands of the government. There’s little transparency and it raises questions about the rule of law.”

OpenAI boss Sam Altman echoed concerns about the government’s choosing of who gets ​access to top models in a post on X.

Extensive safety testing “is not a bad idea. I just don’t like the idea of the government picking the customers,” he wrote.

Experts have said that Mythos models, ​in the wrong hands, could dramatically accelerate sophisticated cyberattacks, particularly in sectors such as banking that rely on complex, interconnected, and often decades-old technology systems.

A letter from U.S. Commerce Secretary Howard Lutnick to Anthropic said there had been “significant ‌progress” in ⁠work done by the company with the government to address “risks associated with the Covered Models.”

It was not immediately clear what safeguards had been adopted. Anthropic said earlier this month that it understood the government believed there is a method of bypassing, or “jailbreaking,” a safeguard that would prevent Fable 5 from being used in identifying software vulnerabilities.

NO EXPORT LICENSE NEEDED FOR NON-US CITIZENS

Lutnick said in the letter that an export license will no longer be needed for Mythos 5 to trusted companies and their employees who are not U.S. citizens, or to Anthropic’s employees who are not U.S. citizens, but licensing restrictions will remain in place for ​companies that are not on the approved list.

The source ​said many of the approved companies are ⁠part of Anthropic’s Project Glasswing, which includes about 100 well-known tech companies and institutions.

The government is also moving towards allowing Anthropic to release Fable soon, although a timeline is unclear, the source said.

Both Fable 5 and Mythos use the same underlying AI model, but Fable 5 is designed to be widely available ​for public use whereas some safeguards are lifted for Mythos.

Both Anthropic and OpenAI plan to go public.

Anthropic’s relationship with the U.S. government has, however, been ​particularly rocky. The company refused ⁠to allow the U.S. military to use its AI models for domestic surveillance and fully autonomous weapons systems and the government retaliated by putting it on a national security blacklist.

The government’s restrictions on Anthropic and OpenAI follow Trump’s signing of an executive order this month establishing a voluntary framework for AI developers to offer “covered frontier models” to the U.S. government for up to 30 days before releasing them to trusted partners.

The administration’s latest ⁠order is “a practical ​interim step, but leaves unresolved the larger issue of how companies can widely release updated models,” said Kate Koren, an ​analyst at the Center for Strategic and International Studies in Washington and a former Commerce Department official.

“The longer there isn’t a system in place that will allow U.S. companies to widely release new models, the more likely it is that China will be ​able to catch up,” she said.