The Night the US Killed Anthropic's Best AI for Everyone

On June 12, 2026, at 5:21 p.m. Eastern, the U.S. Commerce Department's Bureau of Industry and Security (BIS) issued a directive that instantly reshaped the global AI landscape. It ordered Anthropic to suspend all access to its two most advanced AI models—Fable 5 and Mythos 5—for any foreign national, anywhere in the world, including Anthropic's own foreign-born employees . Because Anthropic had no practical way to verify the nationality of its users in real time, the company was forced to disable both models for every customer on earth . Within hours, two of the world's most powerful AI tools, which had been publicly available for just three days, went dark .

The administration cited national security concerns, specifically the risk that the models' built-in safeguards could be "jailbroken," giving foreign adversaries powerful offensive cyber capabilities . In a letter to Anthropic CEO Dario Amodei, Commerce Secretary Howard Lutnick warned that the company needed an export license from BIS to grant any foreign national access to the models, threatening criminal and civil penalties for non-compliance .

From Hardware to Software: An Unprecedented Shift

For decades, U.S. export controls focused on physical items: advanced semiconductors, chipmaking equipment, and supercomputers. This order marked a dramatic shift, for the first time treating the intangible weights and software of a deployed frontier AI model as a controlled, defense-related item . The move was immediately described as a "structural inflection" in AI governance , an "unprecedented" and "sweeping" action . Critics quickly condemned the framework as an "ad hoc licensing regime," where a single-agency letter could kill global access to a cutting-edge technology overnight without congressional debate or any semblance of due process .

The Global Revolt: AI Sovereignty Goes Mainstream

The directive turned an abstract policy fear into a live demonstration of dependency. Overnight, the debate over AI sovereignty moved from think tanks to the highest levels of government.

• Canada's immediate response: Prime Minister Mark Carney seized on the order just two days later, arguing it was proof that Canada "must build its own AI infrastructure rather than depend on US-controlled platforms" . The ban was framed as a direct threat to a middle power's security and autonomy .
• Europe sounds the alarm: In France, Le Monde ran the headline "'The AI war has begun'" . The European Commission stated that such measures "should not be discriminatory against partners" and saw the incident as another pressing reason to achieve technological autonomy .
• A collective scramble: Governments from the UK, Japan, and India began reassessing their reliance on US-headquartered AI providers, pushing for national AI infrastructure investments to ensure they couldn't be disconnected at Washington's whim .

The core question, now impossible to ignore, was stark: could allied nations trust a U.S. licensing regime that could revoke access to frontier AI at any moment, for any reason, with no reciprocal control?

Cohere's Moment: A Sovereign Alternative Drowns in Inbounds

The immediate commercial beneficiary was Toronto-based Cohere, a company that has long positioned itself as a sovereign, non-US alternative by offering on-premises AI deployment . Within days of the Anthropic order, Cohere’s Chief AI Officer Joelle Pineau reported a "huge number of inbounds" from governments and investors suddenly seeking a model that runs inside their own data centers, under their own legal jurisdiction, and out of reach of a U.S. government kill switch .

This surge confirmed Cohere’s pre-existing strategy. Just weeks earlier, in April 2026, the company had acquired German AI lab Aleph Alpha, creating a combined entity valued at $20 billion that was endorsed by both the Canadian and German governments . The Anthropic export ban transformed Cohere’s business pitch from a theoretical differentiator into an urgent procurement priority for nations spooked by the sudden loss of access to U.S. models .

An Own Goal? The Broader Fallout

Four days after the order, the strategy of protecting U.S. AI leadership appeared to be backfiring spectacularly . As Cohere’s inbound inquiries surged, China’s DeepSeek closed a record $7.4 billion funding round, and Chinese AI labs began slashing their token prices by up to 99% . The export control, designed to contain foreign access to advanced AI, was instead accelerating demand for non-US alternatives—including those from U.S. competitors—on a global scale .