At the G7 summit in Evian-les-Bains, France (beginning June 15, 2026), French President Emmanuel Macron made an urgent call for the world's wealthy democracies to cooperate on regulating advanced AI systems and urged the United States not to withhold frontier AI technology from allies. The appeal came days after a Trump administration directive ordered Anthropic to take its two newest models offline to prevent foreign national access — a move Macron characterized as a "strictly nationalist reaction" with serious geopolitical consequences.
Context: The US Model Access Restriction
On June 12, 2026, the US government directed Anthropic to suspend access to its newest frontier models — Fable 5 and Mythos 5 — citing national security concerns about access by foreign nationals. While Macron acknowledged it was "a good thing" that US officials recognized the potential dangers of the most capable AI systems, he warned that unilateral access restrictions without allied consultation risked undermining the commercial value of US AI companies and eroding trust among democratic partners.
The underlying anxiety is strategic: if allied nations cannot reliably access leading-edge AI models from US providers, they face pressure to develop or source alternatives — potentially from less transparent suppliers or, in a worst case, from authoritarian competitors.
Macron's Positions at the G7
Convening a working AI lunch with top lab leaders — including OpenAI CEO Sam Altman, Google DeepMind CEO Demis Hassabis, Anthropic CEO Dario Amodei, and representatives from France's Mistral, Germany's Black Forest Labs, and Japan's Sakana AI — Macron outlined several interlinked positions:
- Allied AI sharing: Democratic nations should cooperate on frontier AI access to ensure authoritarian regimes do not gain relative advantage
- Formalized AI security cooperation: Government-to-government frameworks should be established for AI safety and cybersecurity coordination, analogous to existing intelligence-sharing arrangements
- Domestic AI investment: France would increase sovereign AI funding as an insurance policy against breakdown of international cooperation
- Governance urgency: The G7 should move toward establishing binding or quasi-binding structures before frontier AI capabilities outpace any regulatory framework
Industry Response
OpenAI's Sam Altman echoed the call for an "international forum" on AI guardrails, stressing that AI safety governance should not be left solely to technology companies — a notable statement from the CEO of the world's most prominent AI lab. Cohere CEO Aidan Gomez noted broad consensus among G7 participants that "we need something" in terms of formal governance architecture.
Anthropic's Dario Amodei, whose company's models were directly affected by the US access restriction, attended the summit but did not make public statements about the specific directive.
Security Implications
From a cybersecurity and national security perspective, the debate surfaces several critical questions the industry is watching:
AI proliferation risk: The same frontier capabilities that make AI useful for productivity and research also enable novel attack vector development, social engineering at scale, and vulnerability discovery. Who controls access to the most powerful models has direct bearing on the threat landscape.
Allied cooperation vs. sovereignty: Frameworks like the Five Eyes have enabled decades of intelligence sharing while managing classification and need-to-know. Whether analogous structures can be built for AI model access — and on what timeline — remains an open question.
Authoritarian access: Both the US and European positions share the goal of preventing adversarial states from accessing the most capable AI systems. The disagreement is about method: unilateral restriction versus coordinated allied governance.
The G7 communiqué is expected to include language on AI governance cooperation, though binding mechanisms are unlikely to emerge from this summit cycle.