European_AI_Forum_Press_Release_AI_Sovereignty_Summit

EUROPEAN AI FORUM

NO SURPRISE, NO WAKE-UP CALL - THIS IS THE RECKONING

Europe ignored the warnings. Now the European AI Forum calls for an immediate AI Sovereignty Summit.

European AI Forum demands immediate AI Sovereignty Summit

The recent directive by the US government to block foreign nationals from accessing Anthropic’s most advanced models, Fable 5 and Mythos 5, serves as a stark reminder of Europe’s urgent need for technological autonomy.

Europe is not developing frontier models at the scale we need. Europe is not building AI infrastructure at the speed required. Europe relies on US providers and hopes it will be fine.

It is not fine. For years, the European AI Forum has been warning of exactly this scenario. What some now call a wake-up call is in fact the reckoning for years of hesitation. The warnings were loud and clear. They were not heard.

What 13 June 2026 means

For the first time in history, a US government has treated AI models as export-controlled goods, comparable to weapons or strategic raw materials. With a single administrative order, European authorities, companies, and research institutions were cut off from critical AI capabilities. No warning. No right of appeal. No European veto.

Because US authorities did not publish the content of the decision, identify a specific threat, or explain what model capabilities triggered the intervention, we observe a clearly concerning trend. American authorities can cut off access to state-of-the-art AI models for millions of people with a single administrative signature, without specific legislation and without public justification.

Last year, the European debate focused on chip export controls. Today, authorities target artificial intelligence models themselves for state control. If similar restrictions are applied to subsequent models, access to leading AI systems will no longer depend on market capabilities - it will depend on citizenship.

This situation also presents a massive problem for research teams across Europe. Immigrants and dual citizens constitute a significant portion of the talent developing AI worldwide. New restrictions hinder collaboration even within American companies and will compel scientists to relocate. The most alarming fact: such serious decisions are triggered in a completely opaque manner, with a general argument about national security deemed sufficient.

Anyone operating critical infrastructure, government agencies, or security-relevant applications based on US AI models must understand one thing: these systems can be shut down at any moment. Without justification. Without warning. Without recourse. This applies to hospitals, energy providers, public administrations, the judiciary, and banks.

When a foreign administration can revoke access to cutting-edge technology globally without a transparent statutory process, European researchers, businesses, and citizens remain exposed. Representing over 3,000 AI companies across the continent, the European AI Forum urges policymakers and industry leaders to redouble their efforts, not in drafting more strategy papers, but in building European alternatives.

Furthermore, we strongly recommend that the European Commission take immediate diplomatic action to urge the US administration to reconsider this stance, bearing in mind that the European Union is a steadfast ally.

What is needed now

The European AI Forum calls on the European Commission and EU member states to immediately convene an AI Sovereignty Summit, bringing together representatives from:

• European Commission
• EU member states
• AI industry and model developers
• Traditional industry sectors
• Security authorities
• Critical infrastructure operators

The goal is a binding action plan for European frontier models with concrete milestones and funding commitments.

Specifically, the European AI Forum demands:

1. Tripling of public investment in European AI model development

Europe needs its own high-performance AI models - open, sovereign, and under European control. Existing programmes such as OpenEuroLLM and SOOFI must be massively expanded immediately. Our continent has the talent, the regulatory framework, and the innovative spirit to build world-class models. We must now provide the resources.

2. Immediate dependency audit across public authorities and critical infrastructure

All government institutions and critical infrastructure operators must systematically assess their AI dependencies on third countries and establish binding timelines for reduction. We must ensure that our access to vital technologies relies on our own democratic principles rather than unpredictable foreign directives.

3. Preferences for genuinely sovereign solutions

Public procurement should introduce preferences for genuinely sovereign cloud and AI solutions in Europe. We must also define and limit ‘sovereign washing’, as a number of non-European entities would have us believe that merely locating data and servers in Europe or employing Europeans is enough to make a solution sovereign.

4. Entrench a sovereignty obligation for public-sector AI applications

For AI systems in sensitive government areas, data storage, processing, and control must remain entirely within Europe - without reliance on providers subject to unilateral access restrictions. To secure our digital future, Europe must aggressively invest in and adopt home-grown AI ecosystems. > “What we experienced yesterday is not a surprise, it is the reckoning for years of hesitation. The US government has proven what we have long been warning: AI is strategic infrastructure. Those who do not control it will be controlled by it. We do not call for another strategy process. We call for a summit. Now. Europe must prioritise European solutions to guarantee our technological sovereignty and protect our global competitiveness.”

• Daniel Abbou, President of the Board, European AI Forum

Background

An unprecedented decision in the history of technology. The US government invoked national security authorities to issue an export control directive forcing Anthropic to suspend access to its Fable 5 and Mythos 5 models for all foreign nationals. The ban also covers individuals living in the United States and experts working for Anthropic on these models.

Crucially, US officials did not disclose the legal basis, the content of the directive, or even the exact duration of its enforcement. The administrative basis stems from the Export Administration Regulations and the deemed export rule, which classifies providing a foreign national access to technology on US soil exactly like exporting that technology abroad.

The public does not know the specific grounds on which officials applied this mechanism. Theoretically, the reason involves potential jailbreak methods. Anthropic representatives argue the demonstration revealed only a small number of already known and minor vulnerabilities, and other public models possess similar flaws.

If this represents the new standard for AI regulation, we are currently witnessing the beginning of an entirely new, protectionist technological era.

1 ## Press contact:

Daniel Abbou, President of the Board daniel.abbou@eaiforum.org | www.eaiforum.org

1. About the European AI Forum The European AI Forum AISBL is the pan-European umbrella organisation representing over $3,000 \mathrm{Al}$ companies, researchers, and innovators. Based in Brussels, EAIF brings together 13 national AI associations and clusters from across Europe. EAIF advocates for a competitive, sovereign, and value-driven AI ecosystem in Europe.↩︎