TL;DR
A June 17 G7 AI lunch in Évian-les-Bains put Dario Amodei, Demis Hassabis and Sam Altman across from political leaders days after U.S. export controls forced Anthropic to disable Fable 5 and Mythos 5 worldwide. Europe’s core demand is no longer abstract AI cooperation; it is reliable access, clear rules for trusted allies and protection against sudden shutdowns ordered by a foreign government.
European leaders used a June 17 G7 working lunch in Évian-les-Bains with Dario Amodei, Demis Hassabis and Sam Altman to press for reliable access to frontier AI after a U.S. export-control directive forced Anthropic to disable Fable 5 and Mythos 5 worldwide, turning fears about foreign dependence into an immediate business and public-sector risk.
The AI session placed the leaders of Anthropic, Google DeepMind and OpenAI in a format usually reserved for political power. The table included U.S. President Donald Trump, French President Emmanuel Macron, German Chancellor Friedrich Merz, European Commission President Ursula von der Leyen and U.K. Prime Minister Keir Starmer, alongside other technology executives and representatives from European and allied AI labs.
The immediate trigger was Washington’s June 12 directive restricting foreign-national access to Anthropic’s most capable models. Axios reported that the Commerce Department treated the models as national-security assets and required licenses for foreign access; Anthropic responded by cutting off customer access because it could not reliably verify nationality at API scale. Business Insider reported that later talks between Anthropic and U.S. officials focused on how to judge model-security risks.
At Évian, the three AI leaders offered versions of international coordination. According to the source material and Financial Times reporting, Amodei urged democratic states to avoid splitting into competing AI blocs, Hassabis backed a Western framework for advanced AI, and Altman pushed for an international forum on testing standards. European leaders pressed a more concrete agenda: durable access to leading models, a trusted-partner route for non-U.S. allies, a voice over compute and chip infrastructure, and child-safety protections built into AI services.
Évian and the fallout: what Europe actually wants
For the first time, Amodei, Hassabis, and Altman sat with heads of state — five days after Washington switched Anthropic’s models off worldwide. Europe’s question: can you rely on models a foreign cabinet can shut down by decree?
The dilemma: what Europe wants from the three CEOs, the three can’t deliver — because they don’t hold the switch, Washington does. Macron’s platform is the right answer, but no fix for a decade-old infrastructure gap. The only answer that doesn’t depend on someone else’s goodwill: your own models, your own compute, open weights you can self-host.
Europe Tests Model Dependence
The shutdown gave European companies and public bodies a concrete example of a risk they had discussed for years: critical AI tools can be withdrawn by a foreign government even when customers are allies, not adversaries. If banks, hospitals, schools, software vendors or ministries build services on closed frontier models, access policy becomes part of operational resilience.
The dispute also links AI safety, trade and industrial policy. U.S. officials say controls on the most capable systems can protect national security; European leaders want the best models without placing public services and private investment at the mercy of sudden U.S. action. That tension is likely to strengthen Europe’s case for local compute, domestic model development, AI gigafactories and self-hosted open-weight systems.
Silicone Case for Rabbit R1 AI Device,Full Protective Cover for Anti-Scratch,Skin-Friendly Dirt and Shock Resistant,Gray
- Compatible Model: Designed for Rabbit R1 AI Device
- Full Protection: Scratch, dust, shock, and drop resistant silicone
- Unobstructed Access: Precise cutouts for buttons and ports
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Anthropic Order Set Agenda
The Évian summit ran from June 15 to June 17, 2026, with Macron using part of the leaders’ program for AI. The official theme cited in the source material was the safe, rapid and effective deployment of AI, but the Anthropic order changed the tone of the conversation.
Anthropic’s models were restricted after U.S. officials cited cybersecurity and national-security concerns. Anthropic has argued that the disputed weaknesses were not unique to its systems and that broad model bans could harm defenders as well as commercial users. Tom’s Hardware reported that the order affected foreign nationals globally, including foreign employees in the United States, leaving the company with no narrow technical path for compliance.
The European response at Évian sits alongside a wider sovereignty push. The source material cites a €420 billion package, AI gigafactories and related capacity programs as part of Europe’s answer, but those projects cannot quickly replace the most advanced U.S. models already embedded in services.
“resist the temptation to splinter”
— Dario Amodei, Anthropic CEO
Access Guarantees Still Missing
It is not yet clear whether Washington will narrow or lift the Anthropic restrictions, how a trusted-partner scheme would work, or whether European users would gain legal rights strong enough to prevent another abrupt cutoff. The technical basis for the U.S. action also remains disputed: officials cited national-security risk, while Anthropic and some outside specialists have argued that the model risks were not unique and should be handled through standards rather than a broad shutdown.
It is also unclear whether Amodei, Hassabis and Altman can give Europe what it wants. They can offer access programs, safety testing and policy support, but export controls, licensing rules and sanctions authority remain government powers.
September Talks May Define Access
According to the source material, leaders are expected to meet again in September, while Macron’s proposed AI platform is due within a month. The near-term test is whether allies can agree on a trusted-partner channel for frontier model access while also setting common rules on cybersecurity and child safety.
Anthropic’s talks with U.S. officials will be the first practical signal. If the ban is narrowed, allies may seek a standing process for future model disputes. If it remains in place, European buyers may accelerate plans for sovereign compute, local models and open systems they can host themselves.
Key Questions
What happened at the G7 lunch in Évian?
G7 leaders held an AI-focused working lunch on June 17, 2026 with Dario Amodei, Demis Hassabis, Sam Altman and other technology executives. The meeting became a forum for Europe’s concerns after U.S. controls forced Anthropic to disable its top models worldwide.
Why did Anthropic shut down Fable 5 and Mythos 5?
The U.S. Commerce Department directed Anthropic to restrict access by foreign nationals, citing national-security concerns. Anthropic said it could not reliably separate users by nationality at API scale, so it cut off access more broadly.
What does Europe want from the AI companies?
European leaders want dependable access to frontier models, a trusted-partner system for allies, safeguards against sudden service loss, more say over compute and chip infrastructure, and stronger protections for children and young users.
Can the AI CEOs give those guarantees?
Only partly. The companies can design access programs and support shared testing standards, but the power to impose or lift U.S. export controls belongs to Washington.
What remains undecided after Évian?
The main open questions are whether the Anthropic ban will be changed, whether allies will get privileged access rules, and how Europe will balance U.S. model access with its own sovereignty plans.
Source: Thorsten Meyer AI
