Anthropic Export Controls Shake G7, CEOs Urge Coalition

Anthropic (P-ANTH) export controls forced the company to disable its frontier models and became the focal point at the G7 working lunch on June 17, 2026, where executives urged a U.S.-led AI coalition to coordinate safety, export rules, and continued access.

U.S. Export Directive Shuts Frontier Models

The U.S. Department of Commerce issued an emergency export-control directive covering Anthropic’s frontier models Fable 5 and Mythos 5, citing national-security concerns that the systems could identify and exploit software vulnerabilities. The order barred access to these models for all foreign nationals, including non-U.S. citizens inside the United States and Anthropic’s foreign-born employees, making reliable citizenship screening impracticable.

To comply, Anthropic disabled global access to Fable 5 and Mythos 5, cutting off public users and enterprise customers while other company models remained online. Officials and industry participants described the move as the first known use of U.S. export controls to restrict access to an AI model itself rather than to hardware, marking a new regulatory tool for limiting model availability.

Anthropic has said it disagrees with the directive, believes there is a misunderstanding about the models’ behavior, and is working with U.S. authorities to address concerns and restore access.

G7 Summit Pushes for U.S.-Led AI Coalition Amid Sovereignty Concerns

On June 17, 2026, a G7 working lunch titled “Ensuring a safe, rapid and effective deployment of artificial intelligence” brought together national leaders and chief executives, including Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman, and Google DeepMind CEO Demis Hassabis. The session focused on frontier AI risks, computing infrastructure, sovereignty, and liability.

In a closed-door meeting, Amodei and Hassabis urged the formation of a U.S.-led AI coalition among like-minded democracies to coordinate safety standards, export controls, and access arrangements for frontier models. They presented the proposal as a conceptual framework rather than a formal treaty or organization, warning that regulatory divergence could fragment safety oversight.

European leaders, including European Commission President Ursula von der Leyen, pressed for cooperative arrangements to preserve access to U.S. AI systems while advancing European “tech sovereignty.” They framed the export-control action as a potential U.S. “kill switch” over foreign users’ access, accelerating plans for regional AI infrastructure and legal safeguards.

The episode intensified pressure on Washington and Anthropic. The company is reported to be working with U.S. authorities to resolve concerns and restore access. Meanwhile, lawmakers in Washington have increased efforts to pursue statutory guardrails for cutting-edge AI models.

Anthropic employees have expressed puzzlement and concern about the restrictions, accusing the administration of targeting the company and warning that the measures threaten the firm’s mission and disrupt foreign-born staff who lost access under the foreign-person rules.

The AI session was among several major topics on the summit’s final day, alongside discussions on Ukraine and China’s industrial overcapacity. Leaders debated nonbinding language on safe AI deployment, including child-safety protections, age-verification measures, and tougher safety obligations for chatbot makers. The Anthropic export-control episode also reverberated in Washington, where bipartisan interest in AI regulation has grown.