The Trump administration has lifted export controls on Anthropic’s two most powerful AI models, Claude Fable 5 and Mythos 5, after the company reached an agreement with the Commerce Department, according to a letter obtained by Wired.
The reversal ends a three-week standoff that began on June 12 when the Commerce Department issued a directive prohibiting foreign nationals from accessing the models, citing cybersecurity concerns tied to their advanced capabilities. The restriction forced Anthropic to suspend access for all users globally, including US employees, as the company scrambled to comply with an order that one senior administration official described as the first test case for a new executive order on AI guardrails.
The dispute
The conflict stemmed from a contested claim that Amazon AI experts had successfully jailbroken Fable 5 after its June 9 launch, bypassing safety mechanisms around advanced cybersecurity capabilities. Anthropic disputed the characterization, arguing its own analysis found only minor vulnerabilities in limited scenarios, not a broad defeat of the model’s protections. The company contended that the same jailbreak technique could affect competing models such as OpenAI’s GPT-5.5, which were not subject to comparable restrictions.
Administration officials accused Anthropic of failing to take the concerns seriously, with one senior official telling Fox Business that the company’s “recklessness” in responding to issues triggered the controls. Anthropic CEO Dario Amodei was reportedly unreachable during early discussions, though sources close to the company denied this.
Deal terms
Following weeks of negotiations involving Commerce Secretary Howard Lutnick, the department agreed to lift restrictions on both models. The deal allows Anthropic to resume global access to Fable 5 and Mythos 5 under what the company described as a collaborative framework with the government.
The dispute unfolded against a fraught backdrop for Anthropic. The company had confidentially filed for an IPO in early June at a valuation approaching US$1 trillion (approximately £800 billion), and the export controls threatened to disrupt its commercial momentum and international partnerships.
The broader industry has watched the case closely, as it represents the first time US export controls have targeted AI software rather than just hardware. Cybersecurity experts had argued that restricting access to top-tier American AI models does not weaken adversaries but instead reduces the competitive edge of US companies.
Sources: Trump admin has lifted export controls on Claude Fable 5 and Mythos 5 (CNBC, June 30, 2026); The Trump Administration Is Lifting Its Export Controls on Anthropic’s Mythos and Fable AI Models (Wired, June 30, 2026); Anthropic proposes collaboration to lift restrictions (Crypto Briefing, June 18, 2026)