TL;DR
European leaders pressed Dario Amodei, Demis Hassabis and Sam Altman at a June 17 G7 lunch after a U.S. export-control directive pushed Anthropic to shut off key models worldwide. The meeting exposed a policy problem for Europe: the AI systems used by its firms and public institutions can still be affected by U.S. executive action. No reversal of the directive was reported, and access guarantees remain unresolved.
European leaders used a June 17 G7 working lunch in Évian-les-Bains to press Dario Amodei, Demis Hassabis and Sam Altman over access to frontier AI models, five days after a U.S. Commerce Department export-control directive forced Anthropic to block its most capable systems worldwide.
The directive ordered Anthropic to block Fable 5 and Mythos 5 for any “foreign national,” according to the source account. Because nationality cannot be checked reliably in real time across API use, Anthropic’s practical response was a worldwide cutoff. European businesses and public institutions that had built the models into operations lost access without warning or a grace period, the account says.
The lunch, hosted by French President Emmanuel Macron during the June 15-17 G7 summit, brought the three leading AI chiefs to the same table as heads of state. The U.S. delegation included President Trump and Secretaries Scott Bessent, Howard Lutnick and Marco Rubio. European participants included Macron, German Chancellor Friedrich Merz, European Commission President Ursula von der Leyen and U.K. Prime Minister Keir Starmer.
The confirmed sequence is narrow but consequential: the directive was issued June 12, the AI lunch took place June 17, and no reversal of the ban was reported. The proposals discussed at the table, including allied access rights, shared testing standards and a trusted-partners model, remain proposals rather than binding rules.
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 AI Dependence
The shutdown turned a long-running sovereignty debate into an operating risk. If European agencies, banks, manufacturers or research teams rely on a model that can be cut off by a foreign order, procurement, resilience planning and cloud architecture all change.
Europe’s demands, as described in the source material, center on six points: reliable long-term model access, protection against another sudden cutoff, a trusted-partners scheme for non-U.S. allies, greater European compute and model capacity, influence over where chips and power infrastructure are built, and child and youth safety rules built into deployment.
The tension is that Amodei, Hassabis and Altman can offer cooperation, forums and technical commitments, but they do not control U.S. export policy. The switch in this case sits with Washington, which limits what company-level assurances can deliver.
Access Control Systems: Security, Identity Management and Trust Models
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Shutdown Changed Summit Agenda
The official G7 AI theme focused on safe, rapid and effective deployment. The source account says the real pressure came from the June 12 export-control move, which showed that U.S.-based AI systems can become instruments of state policy even when customers are outside the United States.
Alongside Amodei, Hassabis and Altman, the lunch included other technology leaders such as Salesforce’s Marc Benioff and Meta AI chief Alexandr Wang, as well as allied and European labs including Mistral, Synthesia, Black Forest Labs, Domyn and Sakana AI.
Macron’s wider response includes a 420 billion euro package tied to technological sovereignty, gigafactories and European AI capacity. Those plans could reduce dependency over time, but they do not solve immediate reliance on U.S.-controlled frontier models.
“Democracies should not give in to the temptation to splinter.”
— Dario Amodei, Anthropic
Advanced API Security: OAuth 2.0 and Beyond
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Access Guarantees Still Missing
It is not yet clear whether the U.S. administration would accept limits on future export-control actions, whether AI companies can offer contracts that survive government orders, or whether a trusted-partners scheme would cover both public institutions and private customers.
The scale of disruption in Europe is also not fully detailed in the source material. It names businesses and public institutions as affected users, but does not identify which organizations lost access, how long outages lasted, or whether customers received compensation.
Enterprise AI Observability and Monitoring: Monitoring, Governing Production AI Systems Drift Detection, LLM Monitoring, Agentic AI, Governance, and FinOps … (Enterprise Machine Learning Operations)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
September Talks Loom Next
A Western democracies platform is expected within a month, according to the source account, with leaders due to reconvene in September. The agenda is expected to include trusted-partner access, cyber-defense cooperation against China-linked risks, child safety principles and the placement of compute, chips and power capacity.
For now, the Anthropic ban remains in place. The next test is whether governments can turn summit language into enforceable access arrangements, or whether Europe moves faster toward self-hosted models, local compute and open-weight systems it can control directly.
trusted partner AI testing platform
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What happened at the G7 AI lunch in Évian?
G7 leaders met top AI executives, including Dario Amodei, Demis Hassabis and Sam Altman, to discuss AI deployment, access and governance after a U.S. directive forced Anthropic to block key models worldwide.
Why did the Anthropic shutdown affect Europe?
The directive applied to any “foreign national.” The source account says Anthropic could not reliably check nationality at API scale, so European users lost access when the company shut the models down globally.
What does Europe want from AI companies?
European leaders want durable access to frontier models, protection from sudden cutoffs, a trusted-partners scheme, more say over AI infrastructure, stronger local capacity and child safety protections.
Did the summit reverse the U.S. directive?
No reversal was reported. The meeting produced proposals and political pressure, but the Commerce Department restriction remained in place in the source account.
Can the AI chiefs give Europe the guarantees it wants?
Only partly. The companies can support access frameworks and technical standards, but U.S. export-control authority belongs to Washington, not to Anthropic, Google DeepMind or OpenAI.
Source: Thorsten Meyer AI
