issued Saturday, June 27, 2026 · 09:16 AM — Updated, Saturday, June 27, 2026 · 01:29 PM
[WASHINGTON] Anthropic reported on Friday (June 26) that the U.S. government has allowed its powerful Claude Mythos 5 artificial intelligence model to be made available to some “trusted” U.S. organizations, partially reversing a two-week-old order halting access citing national security risks.
More than 100 companies and institutions, including many Fortune 500 companies, will have access to Mythos 5, according to people familiar with the new directive, who asked not to be named due to the sensitivity of the matter.
Concerns that powerful AI systems could be misused by military intelligence users in China, Russia, or other countries of concern have led President Donald Trump’s administration to take an aggressive approach in monitoring the release of Anthropic and competing OpenAI frontier models.
OpenAI announced on the same day that it would postpone the full release of GPT-5.6 at the request of the US government, restricting access to a small group of vetted partners whose details have been shared with authorities.
Anthropic abruptly disabled its latest AI models, Mythos 5 and Fable 5, for all users following a June 12 government export control order.
“Today, the government notified us that our most powerful cybersecurity model, Mythos 5, can be reintroduced to a range of U.S. organizations that operate and defend critical infrastructure,” Anthropic said in a statement Friday.
“We are quickly restoring access to these organizations and continue to work with governments to expand access to Mythos 5 and make Fable 5 generally available again.”
Governments are criticized for choosing who has access
The government’s review of which companies can access Mythos has drawn much criticism.
“No one knows how these companies are chosen or why others are excluded,” said John Coleman, legislative counsel for the Foundation for Individual Rights and Expression, a nonpartisan free speech organization based in Philadelphia.
“This puts too much power in the hands of the government, with little transparency and raises questions about the rule of law.”
In a post about X, OpenAI boss Sam Altman echoed concerns about governments choosing who has access to top models.
Massive safety testing “isn’t a bad idea; I just don’t like the idea of the government choosing its customers,” he wrote.
In the wrong hands, experts say the Mythos model could dramatically accelerate sophisticated cyberattacks, especially in sectors such as banking that rely on complex, interconnected, and often decades-old technology systems.
A letter from U.S. Secretary of Commerce Howard Lutnick to Anthropic said the company had made “significant progress” in its work with the government to address “risks associated with the targeted models.”
It was not immediately clear what safety measures were taken. Anthropic said earlier this month that it understands the government believes there is a way to circumvent, or “jailbreak”, safeguards that prevent Fable 5 from being used to identify software vulnerabilities.
No export license is required for non-US nationals
Lutnick said in the letter that Mythos 5 export licenses will no longer be required for trusted companies and their employees who are not U.S. citizens, or for Anthropic employees who are not U.S. citizens, but licensing restrictions for companies not on the approved list will remain.
Many of the approved companies are part of Anthropic’s Project Glasswing, which includes about 100 prominent technology companies and institutions, the person said.
The government is also moving toward allowing Anthropic to release “Fable” soon, but the timeline is unclear, the people said.
Both Fable 5 and Mythos use the same underlying AI model, but Fable 5 is designed to be widely available to the public, while Mythos has some safeguards lifted.
Anthropic and OpenAI both plan to go public.
But Anthropic’s relationship with the U.S. government was particularly volatile. The company refused to allow the U.S. military to use its AI models for domestic surveillance or fully autonomous weapons systems, and the government retaliated by placing the company on a national security blacklist.
The government restrictions on Anthropic and OpenAI come after President Trump signed an executive order this month establishing a voluntary framework for AI developers to provide “covered frontier models” to the U.S. government for up to 30 days before releasing them to trusted partners.
Kate Koren, an analyst at the Center for Strategic and International Studies in Washington and a former Commerce Department official, said the administration’s latest order “is effectively a temporary measure, but it leaves open the big question of how companies can widely release their latest models.”
“The longer U.S. companies are not able to widely launch new models, the more likely it is that China will catch up,” he said. Reuters
