The U.S. government has granted Anthropic permission to restore access to its powerful Claude Mythos 5 artificial intelligence model for a select group of American organizations, offering a partial rollback of sweeping export controls imposed two weeks ago amid national security concerns over potential misuse by foreign adversaries.
The decision allows more than 100 companies and institutions, including many Fortune 500 firms, to once again use the model, which is particularly noted for its advanced capabilities in identifying software vulnerabilities. The move was confirmed by a source familiar with the directive, who spoke to Reuters on condition of anonymity due to the sensitivity of ongoing discussions between the company and officials in Washington.
“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.”
Register for Tekedia Mini-MBA edition 20 (June 8 – Sept 5, 2026).
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
The partial reversal comes after Anthropic had disabled both Mythos 5 and Fable 5 for all users following a Commerce Department export control order on June 12. That directive had cited risks that the models could be exploited by military or intelligence entities in countries of concern, including China and Russia.
Commerce Secretary Howard Lutnick’s letter to Anthropic highlighted “significant progress” in collaborative efforts to mitigate those risks. Under the updated guidance, no export license is now required for Mythos 5 when provided to approved U.S. organizations and their non-U.S. citizen employees, nor for Anthropic’s own non-U.S. staff. Licensing restrictions, however, remain in place for any entities not included on the approved list.
Many of the newly authorized organizations are participants in Anthropic’s Project Glasswing, a cybersecurity initiative involving around 100 prominent tech companies and institutions.
A Fragile Truce in the AI Oversight Battle
This development represents a tentative truce in what has been a notably strained relationship between Anthropic and the Trump administration. The company had been placed on a national security blacklist earlier this year after refusing to allow its technology to be used for mass domestic surveillance or fully autonomous weapons. The recent export controls added to the friction, but Friday’s partial lifting suggests negotiations are yielding incremental progress.
The episode underscores the delicate balancing act facing both AI developers and policymakers in Washington. On one side are legitimate worries about powerful vulnerability-discovery tools falling into adversarial hands and accelerating cyberattacks on critical infrastructure. On the other hand, there is the risk that overly restrictive measures could slow American innovation and hand an advantage to international competitors.
OpenAI faced a similar situation earlier on Friday, announcing it would delay the full public launch of its GPT-5.6 models at the government’s request and initially limit access to a small group of vetted partners. Both companies have voiced unease about government vetting becoming a long-term norm.
“We don’t believe this kind of government access process should become the long-term default,” OpenAI said in its statement. “It keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them.”
Critics have been vocal about the lack of transparency in how the government selects which organizations gain access.
“No one knows how these companies are picked and why everyone else is excluded,” said John Coleman, legislative counsel for 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 CEO Sam Altman echoed these sentiments 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 warned that models like Mythos, if misused, could dramatically speed up the discovery and exploitation of vulnerabilities in complex systems, particularly in sectors like banking that rely on legacy infrastructure. A letter from Lutnick noted that an export license will no longer be needed for Mythos 5 to trusted companies and their employees who are not U.S. citizens, but restrictions will persist for non-approved entities.
Tackling the National Security Concerns
The government’s actions follow President Donald Trump’s executive order earlier this month, which established a voluntary framework for AI developers to submit “covered frontier models” for up to 30 days of government review before releasing them to trusted partners. Kate Koren, an analyst at the Center for Strategic and International Studies and a former Commerce Department official, described the latest directive as “a practical interim step, but leaves unresolved the larger issue of how companies can widely release updated models.”
“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 warned.
The situation with both Anthropic and OpenAI underpins that national security considerations are increasingly shaping the pace and direction of AI development in the United States. As frontier models grow more capable, the tension between rapid innovation and precautionary oversight is only likely to intensify.
Analysts note that the partial restoration of access to Mythos 5 provides some breathing room for Anthropic, allowing it to support critical infrastructure defenders while it works toward broader availability. The company’s plans to go public add another dimension, as greater transparency and regulatory engagement will become even more important.
However, the events of the past few weeks suggest a pragmatic, evolving approach is taking shape — but fundamental questions remain about how sustainable this framework will be as AI capabilities continue to advance. In an industry where the line between breakthrough and risk is increasingly blurred, both companies and regulators are navigating uncharted territory.