The US company Anthropic was forced to deactivate access to its most advanced artificial intelligence models, Fable 5 and Mythos 5, for all foreign users, including its own international employees, following an order from the United States government. This event has reignited the debate on digital sovereignty and the global control of AI technologies.
What happened
On June 12, 2026, Anthropic, a leading company in frontier AI, received an export control directive from the United States. The order mandated the company to immediately suspend access to its flagship models, Fable 5 and Mythos 5, released just days earlier on June 9, to all non-US nationals The Verge AI. Anthropic had stated that Fable 5's capabilities exceeded those of any previous model, making the shutdown particularly significant The Verge AI. The decision, communicated amidst national celebrations, caught the company by surprise, forcing it to act quickly to comply, blocking access even for many of its own researchers and engineers operating outside the United States The Verge AI. This episode highlighted how US tech companies, despite operating on a global scale, are subject to their government's directives, with implications extending far beyond national borders.
Why it matters
The US government's intervention with Anthropic brings to light a crucial question: who controls the most advanced AI technologies, and what are the implications for international AI governance? The incident serves as a stark reminder for non-US countries, who face the potential vulnerability of their infrastructures and innovation capabilities if they rely on AI models developed and controlled by the US. This scenario strengthens the argument for digital sovereignty, pushing nations to invest in developing their own artificial intelligence capabilities to avoid dependencies that could be severed by external political decisions. The sudden block not only hinders global research and development but also raises questions about trust and stability in international technological collaborations. The ability of a single government to deactivate access to frontier AI tools for national security reasons, or other not always transparent motives, can have a profound impact on economic competitiveness, scientific research, and countries' ability to address complex challenges with the aid of AI.
The HDAI perspective
The episode involving Anthropic and the US government's AI model blockade underscores the urgency of establishing robust and multilateral AI governance. For Human Driven AI, this is not merely a technical or commercial problem, but a deeply ethical and social impact issue. Our vision of ethical AI is intrinsically linked to transparency, fairness, and international collaboration. A government's unilateral decision to restrict access to key technologies risks creating an even wider digital divide, penalizing less developed regions and hindering the resolution of global problems that require a collaborative approach. AI governance cannot be left to unilateral decisions that ignore global implications and the need for fair and responsible access. It is crucial that the dialogue on AI regulation shifts towards a more inclusive model, considering the needs and perspectives of all nations, promoting shared ethical standards and independent audit mechanisms. Topics like this will be central to the HDAI Summit 2026 in Pompeii, where global experts will gather to outline a future for AI that truly serves humanity.
What to watch
The reaction of the international community and the future moves of European governments and other global powers will be crucial. We anticipate an acceleration of efforts to create national AI infrastructures and the development of "sovereign" models, less dependent on US big tech. The application of the EU AI Act could become a model for regulation that balances innovation and control, providing a framework for digital autonomy. It will be interesting to observe whether this event prompts greater cooperation among countries to establish international standards and protocols for the export and use of frontier AI technologies, thus preventing technology from becoming a tool of division rather than shared progress.