Washington Lifts Export Controls on Anthropic's Claude Fable 5 and Mythos 5 After Safety Negotiations

A Sudden Halt and a Diplomatic Resolution

The Trump administration has officially lifted export controls on Anthropic's most advanced artificial intelligence models, Claude Fable 5 and Mythos 5, resolving a high-stakes standoff that had frozen the company's frontier technology for weeks. In a letter sent to Anthropic co-founder Tom Brown, Commerce Secretary Howard Lutnick confirmed that a license is no longer required for the export, reexport, or in-country transfer, including deemed export or deemed reexport, of the Mythos or Fable models. The decision marks the end of a dispute that began in mid-June 2026, when the Commerce Department issued an export ban that forced Anthropic to suspend model access for all foreign nationals, including its own non-U.S. employees.

The regulatory freeze was originally triggered after Amazon, acting as a trusted partner, discovered a jailbreak vulnerability that bypassed the safety guardrails on Fable 5. Fable 5 is a version of Anthropic's highly powerful Mythos model designed with additional safety restrictions for public deployment. Although Anthropic initially argued that the vulnerability was relatively simple, common among other public models, and that ensuring zero jailbreaks was an impossible standard that would halt industry-wide innovation, the government's national security concerns prevailed. This led Anthropic to completely disable customer access to both models to comply with the federal order.

The Strategy Shift: Personnel and Safeguards

To resolve the impasse, Anthropic underwent a significant tactical and diplomatic shift in its dealings with the White House and the Commerce Department. Commerce Secretary Lutnick and National Cyber Director Sean Cairncross led the administration's negotiations. According to sources familiar with the matter, Anthropic CEO Dario Amodei was replaced in negotiation meetings by co-founder Tom Brown, whom government officials reportedly found more personally agreeable. Along with this personnel change, Anthropic adjusted its communication style, moving away from debating the theoretical impossibility of stopping all jailbreaks and instead focusing on building more robust safeguards to satisfy the administration's demands.

As part of the final agreement, Anthropic committed to proactively detecting and addressing security risks associated with its models. The company also agreed to work closely with the federal government to establish protocols, standards, and release schedules for Mythos, Fable, and future systems. To directly address the specific vulnerability exposed by Amazon, Anthropic implemented a new safeguard designed to block the reported exploit, paving the way for the administration's approval.

Restoring Access and Pivoting to New Horizons

Following the resolution, Lutnick announced on social media that his department had worked closely with Anthropic over a two-week period to analyze and approve Fable 5, aiming to align government standards and strengthen American leadership in AI. Under the new guidelines, Anthropic has begun restoring access to its systems. Fable 5 is available to global users on the company's cloud platform starting today, while access to the more restricted Mythos 5 model has been restored for select U.S. organizations. Prior to the full lifting of controls, the Commerce Department had allowed a limited release of Mythos 5 to a group of key partners, known as Project Glasswing, to secure critical software against rapid cyber exploits.

The resolution of the export dispute arrives at a critical moment for the company and the broader industry. While Anthropic worked to restore its flagship models to the market, competitors have continued to advance, with reports of a new Chinese model that allegedly performs as well as, if not better than, Fable and Mythos. Amid these market pressures, Anthropic is also expanding its operational footprint, with the company's head of life sciences recently announcing the launch of an internal drug discovery program to build specialized AI tools for drug manufacturers. Whether this resolution signals a sustainable framework for frontier AI regulation or merely demonstrates that corporate compliance and personnel changes can bypass national security freezes remains the critical question for an industry increasingly caught in the crosshairs of geopolitical trade policy.