Ordered by Washington to restrict access to its most advanced models for foreign nationals, Anthropic deactivated them for all of its customers. The first known application of export controls to a large language model, the episode turned into a public standoff and reignited, in Europe, the debate over technological dependence.
It is an unprecedented event in the recent history of artificial intelligence. On Friday, June 12, 2026, at 5:21 p.m. ET, Anthropic received an export directive from the U.S. Department of Commerce ordering it to suspend access to its two most advanced models - Fable 5 and Mythos 5 - for any foreign national, whether located outside the United States or on U.S. soil, including the company’s own non-American employees.
Unable to reliably distinguish its users by nationality, Anthropic chose to deactivate both models outright for all of its customers, Americans included. Amazon Web Services, which hosts them, confirmed that access had been revoked for all users and all regions. The company’s other models, including Claude Opus 4.8, are not affected.
What makes the episode unprecedented
Until now, U.S. export controls had targeted hardware - chips and compute servers - mainly bound for China. This is the first time, to the knowledge of observers, that such a measure has directly hit access to a large language model, and by extension all international customers of an AI provider.
At the origin of the decision: a cyber-related concern. According to Anthropic, the government had learned of a “jailbreak” method for Fable 5 - a workaround that bypassed the safeguards separating the public-facing model from the unrestricted cyber capabilities of Mythos, the model it is derived from. The company says it reviewed the demonstration and found only narrow, already known vulnerabilities, which it says are also present in other public models that were not targeted.
A standoff, two versions
Initially presented as a unilateral government action, the affair has turned into a standoff with conflicting accounts. On June 13, David Sacks, co-chair of the Presidential Advisory Council on Science and Technology and former administration “AI czar,” gave the executive branch’s version: a trusted partner of Anthropic and the government - identified by several media outlets as Amazon - allegedly discovered the jailbreak during testing; the administration then asked Dario Amodei to fix the flaw or remove the model, and the executive reportedly refused. The export control, Sacks wrote, was taken “reluctantly,” and restoration now depends solely on a fix: “the ball is in Anthropic’s court.”
The company disputes that reading. It argues that a narrow workaround did not justify pulling a model used by millions of people and, according to a source cited by Fortune, says it had only 90 minutes to respond, with no prior warning of a national security threat. Behind the scenes, Semafor’s coverage links the decision to concerns about Chinese access to Mythos’s capabilities - placing the episode back within the logic of export controls aimed at China, rather than a simple product-security dispute.
As of June 15, access has still not been restored. The case comes at a sensitive time for Anthropic, which filed a confidential IPO prospectus in early June valuing the company at around $965 billion; regulatory risk is now part of the listing narrative.
Europe reacts, from left to right
In Europe, the episode triggered a wave of political reactions that cut across party lines, compiled notably by Euronews. In France, Bruno Retailleau (former Interior Minister, candidate in the 2027 presidential race) sees it as a wake-up call: a nation that depends on others for its technology can be “unplugged overnight”; he calls for “rearming” national technological power by building on Mistral, OVHcloud, Scaleway or ChapsVision. Benjamin Haddad, Minister Delegate for Europe, believes Europe cannot be content to be “an open market dependent on technologies designed, financed and controlled” elsewhere. Édouard Philippe (former Prime Minister, mayor of Le Havre) describes AI as critical infrastructure “as essential as electricity or Internet.” Jordan Bardella (president of Rassemblement National, Member of the European Parliament) calls for accelerating support for “the Mistral AI gem.”
The conclusion goes beyond France. In the Netherlands, Geert Wilders links the affair to national sovereignty; in the United Kingdom, MPs Al Carns and Tom Tugendhat, both former ministers, note that British researchers, companies and hospitals had been using the now-cut-off model.
Why this matters for companies
Beyond the Anthropic case, the episode turns a long-theoretical question into a concrete one. As long as access to models was governed solely by commercial contracts, AI sovereignty remained an abstract debate. The directive shows that, in the absence of a fallback, a European organization can lose access to a critical tool overnight, by decision of a foreign administration, with no notice and no immediate recourse.
The right framing is not “should we leave U.S. providers?” - that would be simplistic - but rather: which critical uses can withstand an interruption, restriction, or unilateral change in access? The episode fits into a broader trajectory: eleven days earlier, a presidential decree had established a government review of the cyber capabilities of “frontier” models; and as early as 2025, an initial regulatory attempt (the “AI Diffusion Rule,” since repealed) had considered controlling not only chips, but also certain model weights. Frontier AI is entering the realm of strategic assets, on par with semiconductors.
To go further To place this episode in the broader context of dependencies on models, cloud, chips and export controls, IntelligenceArtificielle.com publishes a reference note on the new dependencies of artificial intelligence - mapping of critical layers, European alternatives, China’s role and a risk framework for decision-makers.
Primary sources: Anthropic statement and technical status update (directive received on June 12, 2026); Reuters, Fortune, Time, CNBC, Semafor, BleepingComputer, Tom's Hardware; David Sacks post (June 13, 2026); Euronews, “Wake-up call: Europe reacts to Anthropic halting access to its Fable 5 and Mythos 5 AI models” (June 13, 2026).
Analysis and context: IntelligenceArtificielle.com reference note on the new dependencies of artificial intelligence (mapping of critical layers, European alternatives, China’s role, risk framework).