By Kashmira Konduparty
The Trump administration has lifted export controls on Anthropic’s most advanced artificial intelligence models after the AI startup agreed to implement additional safeguards addressing national security concerns, the company said Tuesday. The decision reverses restrictions imposed less than three weeks ago that had blocked foreign nationals from accessing the company’s newest AI systems.
The U.S. Commerce Department removed the restrictions on Anthropic’s Claude Fable 5 and Mythos 5 models after the company worked with federal officials to strengthen protections against misuse. Anthropic said it will begin restoring access to the models across its Claude platform, Claude Code, Amazon Web Services, Google Cloud and Microsoft Foundry.
READ: US restricts foreign access to Anthropic AI models (June 14, 2026)
The administration imposed the export controls on June 12, citing national security concerns over the potential misuse of the models. The government was particularly concerned that Fable 5 could be “jailbroken,” or manipulated into identifying software vulnerabilities that could be exploited in cyberattacks. Anthropic said at the time that it disagreed with the order but complied immediately by disabling access for foreign nationals worldwide, including some of its own employees.
Commerce officials blocked foreign nationals the restrictions were lifted after Anthropic implemented stronger security measures and agreed to work more closely with the government on AI safety standards. The company said Fable 5 now incorporates its strongest safeguards to date, including systems designed to detect and block attempts to bypass safety protections. Anthropic also agreed to report malicious activity and collaborate on broader industry standards for frontier AI security.
The dispute marked one of the most significant instances of U.S. export controls being applied to advanced AI models rather than computer chips or other hardware. It also highlighted growing concerns within the federal government over the potential military and cybersecurity implications of increasingly capable artificial intelligence systems.
READ: Anthropic urges US to not block state AI regulations till federal standards are set (June 11, 2026)
The restrictions had disrupted businesses and software developers that rely on Anthropic’s AI products, while raising questions about how the United States intends to regulate cutting-edge AI technologies. Industry observers noted that rival companies continued offering comparable models during the suspension, prompting debate over how export controls should be applied across the sector.
Anthropic said Mythos 5, which is designed primarily for advanced cybersecurity applications, will continue to be rolled out in phases, with priority given to trusted organizations in the United States before broader availability. The company acknowledged that no AI safeguard is foolproof but said the updated protections significantly reduce the likelihood of successful jailbreak attempts.
The reversal represents a significant policy shift for the Trump administration as it seeks to balance national security concerns with maintaining U.S. leadership in artificial intelligence development.