Trump administration allows Anthropic to release Mythos AI to ‘trusted’ US organizations
Listen to this article
Average 5 minutes
The audio version of this article was created by AI-based technology. It can be mispronounced. We are working with our partners to continuously review and improve the results.
Anthropic said on Friday that the US government had allowed it to release its powerful Claude Mythos 5 intelligence model to some “trusted” American organizations, partially reversing an order two weeks ago to freeze access due to national security risks.
More than 100 companies and institutions will now be able to receive Mythos 5, including several Fortune 500 companies, said a source familiar with the new order, who declined to be identified due to the sensitivity of the matter.
Concerns that powerful AI systems could be misused by military intelligence users in China, Russia or other worrisome countries have prompted the administration of US President Donald Trump to take an aggressive approach to overseeing the release of Anthropic frontier models and rival OpenAI.
OpenAI said earlier in the day it was delaying the full public launch of GPT-5.6 at the request of the US government, limiting its access to a small group of vetted partners whose information has been shared with authorities.
Anthropic had suddenly 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 informed us that the Mythos 5, our most robust cybersecurity model, can be deployed again to a group of US organizations that operate and protect critical infrastructure,” Anthropic said in a statement on Friday.
“We are restoring access to these organizations immediately, and we continue to work with the government to expand access to Mythos 5 and make Fable 5 available for general use again,” it added.
In the meantime9:44How powerful is the Anthropic Mythos?
Anthropic didn’t release its latest AI model “Mythos” to the public, but only released a composite of 40 companies because it says it’s the most powerful when it comes to cybersecurity. It has found bugs in some of the world’s most protected systems, and if Mythos falls into the wrong hands, it could leave hundreds of organizations at risk. Lily Hay Newman, senior writer at WIRED opens up about it all.
The government is faced with the dilemma of choosing who has access
The government’s assessment of which companies cannot access Mythos has been widely criticized.
“No one knows how these companies are chosen and why everyone else is being shut out,” said John Coleman, legal counsel for the Philadelphia-based free speech organization the Foundation for Individual Rights and Expression.
“This puts a lot of power in the hands of the government. There is little transparency and it raises questions about the law.”
OpenAI CEO Sam Altman has voiced concerns about the government’s choice of who gets the top models in the X post.
Broader security checks are “not a bad idea. I just don’t like the idea of the government picking on customers,” he wrote.
Experts said Mythos models, in the wrong hands, can greatly accelerate cyberattacks, especially in sectors such as banks that rely on complex, interconnected, and often decades-old technology systems.
A letter from US Commerce Secretary Howard Lutnick to Anthropic said there has been “significant progress” in the work done by the company and the government to address “risks associated with Integrated Models.”
It was not immediately clear what precautions had been taken. Anthropic said earlier this month that it understands the government believes there is a way to bypass, or “jailbreak,” security that would prevent Fable 5 from being used to identify software vulnerabilities.
No export license is required for non-US citizens
Lutnick said in the letter that an export license would no longer be required for Mythos 5 for trusted companies and their non-US citizen employees, or for Anthropic’s non-US citizen employees, but licensing restrictions would remain in place for companies not on the whitelist.
The source said many of the approved companies are part of Anthropic’s Project Glasswing, which includes about 100 well-known companies and technology institutes.
The government is also looking at allowing Anthropic to release Fable soon, although the timeline is unclear, the source said.
Both Fable 5 and Mythos use the same basic AI model, but Fable 5 is designed to be widely available for public use and some protections have been proposed for Mythos.
Both Anthropic and OpenAI plan to go public.
In the first open letter from his reign, Pope Leo urged governments to slow down the development of AI systems, saying they prioritize conflict and risk leading the world down a path of endless war.
Anthropic’s relationship with the US government, however, has been very rocky. The company refused to allow the US military to use its AI models for home surveillance and autonomous weapons systems, and the government retaliated by placing it on a national security blacklist.
The government’s restrictions on Anthropic and OpenAI follow Trump’s signing of an executive order this month that establishes a voluntary framework for AI developers to provide “borderline models” to the US government for up to 30 days before releasing them to trusted partners.
The administration’s latest order “is an effective temporary measure, but it leaves unsolved a major problem of how companies can release the revised models,” said Kate Koren, an analyst at the Center for Strategic and International Studies in Washington and a former Commerce Department official.
“If for a long time there is no plan that will allow American companies to widely release new brands, there are many chances that China will be able to find it,” he said.
