BY COMFORT OGBONNA
Anthropic announced Tuesday that the U.S. Department of Commerce has lifted export restrictions on its Fable 5 and Mythos 5 artificial intelligence models, marking a significant reversal less than three weeks after the company was ordered to suspend access to the advanced systems over national security concerns.
The decision comes after Anthropic agreed to strengthen security measures and work more closely with the U.S. government to identify and address risks associated with its most powerful AI technologies.
The company confirmed the policy change in a post on X, saying it had received official notice that export controls on both Fable 5 and Mythos 5 had been withdrawn. Anthropic added that it would begin restoring access to the models the following day, allowing customers to once again use the systems after weeks of restrictions.
The Commerce Department had imposed the export controls on June 12 as part of a broader effort by Washington to monitor the release of cutting-edge artificial intelligence systems. U.S. officials have expressed growing concern that increasingly capable AI models could be exploited by foreign military or intelligence agencies, particularly in countries such as China and Russia, for cyber operations, surveillance, or other activities considered threats to national security.
Anthropic had responded to the restrictions by disabling access to both Mythos 5 and Fable 5. Last week, however, the U.S. government partially eased those measures by allowing Mythos 5 to be used only by a limited group of trusted U.S. organizations while discussions continued over broader access.
According to a letter from U.S. Commerce Secretary Howard Lutnick, the export restrictions were removed after Anthropic committed to proactively identifying and addressing security risks associated with its AI models. The company also agreed to collaborate with federal officials on developing security protocols for Mythos, Fable, and future AI systems, while promptly notifying the government of any malicious attempts to misuse the technology.
Despite lifting the restrictions, Lutnick made clear that the Commerce Department retains the authority to reimpose export controls if circumstances change or if Anthropic fails to uphold the commitments outlined in its agreement with the government.
Although Fable and Mythos are built on the same underlying AI technology, they are designed for different purposes. Fable is intended for broad public availability with extensive safety protections, while Mythos is aimed at specialized users and operates with fewer restrictions to support more advanced applications.
A source familiar with the matter said Anthropic has introduced an additional safeguard specifically designed to block a technique that officials believed could be used to bypass, or “jailbreak,” Fable 5’s built-in safety protections. The source also noted that the vulnerabilities identified through that technique had already been discovered and patched before the government’s concerns were raised.
Artificial intelligence security experts have warned that highly capable models such as Mythos could significantly enhance the ability of cybercriminals or hostile state actors to launch sophisticated cyberattacks if the technology falls into the wrong hands. Critical industries such as banking, financial services, and infrastructure are viewed as particularly vulnerable because they often rely on complex networks and decades-old computer systems.
The easing of restrictions on Anthropic follows another closely watched government intervention involving the AI industry. OpenAI recently delayed the wider public release of its GPT-5.6 model at the request of U.S. officials, limiting initial access to a carefully selected group of vetted partners while additional security reviews were conducted.
Those decisions have sparked criticism from some industry observers, who argue that the government’s process for determining which companies qualify as “trusted” lacks transparency and could create an uneven competitive landscape within the rapidly growing AI sector.
Federal oversight of advanced AI models has intensified this month following President Donald Trump’s executive order establishing a voluntary framework that encourages developers of powerful frontier AI models to provide the U.S. government with early access for security evaluation before broader deployment to trusted partners.
Isaac Harris, executive director of the Frontier Security Institute, said the Commerce Department’s decision to remove the restrictions suggests that the United States is beginning to establish clearer standards for evaluating advanced AI systems developed by American companies.
However, Harris cautioned that major questions remain regarding how the U.S. government will address similarly advanced AI technologies developed overseas, particularly in China, where regulatory standards and safety guardrails may differ significantly.
Anthropic and OpenAI have both confidentially filed paperwork for initial public offerings in the United States, highlighting the enormous commercial potential of the rapidly expanding artificial intelligence industry despite increasing regulatory oversight.
Anthropic’s relationship with the U.S. government has faced several challenges this year. Earlier, the Pentagon classified the company as a supply-chain risk, preventing defense contractors from using Anthropic’s AI systems in military-related work after the company declined to allow its technology to be used for mass domestic surveillance or fully autonomous weapons systems. The latest decision by the Commerce Department, however, signals renewed cooperation between Anthropic and federal authorities as both sides seek to balance technological innovation with national security.