The US government has partially lifted its two-week-old suspension of access to Anthropic’s most powerful AI models, according to Reuters.
This partial reversal now allows more than 100 companies and institutions, including many Fortune 500 companies, to regain access to Claude Mythos 5, though the model remains restricted from general public use.
Reuters reports that Anthropic disclosed the development on Friday, saying the government had notified it that Mythos 5 could be redeployed to a set of US organizations that operate and defend critical infrastructure.
The company said it is restoring access quickly and continuing to work with the government to expand Mythos 5 access further and return Fable 5 to general availability.
Anthropic framed the partial reinstatement as progress in its ongoing engagement with the government over the conditions under which its most capable models can be safely deployed.
- “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,” the company said, according to Reuters.
- “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.
A letter from US Commerce Secretary Howard Lutnick to Anthropic, cited by Reuters, said there had been “significant progress” in work done by the company with the government to address “risks associated with the Covered Models”.
However, it was not immediately clear what specific safeguards had been adopted as part of the agreement.
The government’s approach of selecting which companies gain access to frontier AI models has drawn sharp criticism from free speech advocates and industry leaders, who argue the process lacks transparency and concentrates too much power in the hands of federal authorities, Reuters reports.
John Coleman, legislative counsel at the Foundation for Individual Rights and Expression, said the vetting process raises fundamental questions about accountability.
- “No one knows how these companies are picked and why everyone else is excluded. 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,” he said.
OpenAI CEO Sam Altman echoed those concerns in a post on X, saying he does not object to safety testing in principle but takes issue with the government determining who gets access to the most powerful models.
- “Extensive safety testing is not a bad idea. I just don’t like the idea of the government picking the customers,” he wrote.
The partial reversal of the US government’s block of Anthropic’s AI models comes on the same day OpenAI said it was delaying the full public launch of its GPT-5.6 model at the US government’s request.
- Nairametrics had earlier reported that the U.S. government has urged OpenAI to restrict access to its upcoming GPT-5.6 model, limiting availability to a select group of government-approved partners because of the model’s advanced capabilities, according to CNN, which cited a source familiar with the matter.
- The request follows the Trump administration’s export control action involving Anthropic, which resulted in the company restricting public access to its most advanced models, Mythos and Fable, over concerns about their potential cybersecurity applications.
According to the source, both OpenAI and the administration consider GPT-5.6 to be comparable in capability to Anthropic’s Mythos model.
