The irony is almost too rich to ignore. Dario Amodei and Demis Hassabis—the CEOs of Anthropic and Google DeepMind—walked into a closed-door lunch at the G7 summit in France and asked the world to form a U.S.-led AI coalition. Days earlier, Washington suspended access to Anthropic's two newest models, Fable 5 and Mythos 5, citing national security and export control violations. You don't need a crystal ball to see the tension: the people calling for global rules are the same ones being punished by the regulator they want to lead.
Let's be clear about what was proposed. On the final day of the G7, June 17, 2026, Amodei and Hassabis formally called for an international coalition to set global standards and rules for artificial intelligence. OpenAI's Sam Altman was in the room. U.S. President Donald Trump and European leaders listened. The pitch was that AI governance needs to align with American leadership—not fragmented national regulations. But the timing couldn't be more awkward.
Just days earlier, Anthropic voluntarily disabled Fable 5 and Mythos 5 after the U.S. applied export controls. That move looked like a preemptive concession—an attempt to de-escalate before the G7 even started. Meanwhile, the company's AI safety lead resigned in June 2026, warning that the world is 'in peril' and that internal pressures were compromising safety work. Anthropic denied calling for a pause on AI development, saying instead that competitors should have mechanisms that *could* allow for a pause. The mixed messages are deafening.
And then there's the lawsuit. Anthropic is fighting the U.S. Department of Defense over a 'supply chain risk' label that could cost the company up to $5 billion in lost revenue. In a rare show of solidarity, 37 employees from OpenAI and Google DeepMind—led by legendary engineer Jeff Dean—filed an amicus brief supporting Anthropic. So while the CEOs propose a coalition, their own troops are rallying against the U.S. government. That's not exactly a unified front.
The subtext here is a power struggle. European leaders reportedly didn't want to confront Trump directly during the lunch, despite Washington's unilateral suspension of Anthropic's models. That suggests the coalition proposal may get a lukewarm reception from EU nations that favor a more multilateral approach. The U.S. might end up leading a coalition that nobody else truly trusts.
So who wins and who loses? If the coalition happens, Big Tech gets a seat at the rule-making table—and a way to preempt more aggressive regulation. If it stalls, we're back to a patchwork of national rules, with companies like Anthropic caught between safety rhetoric and realpolitik. The average person should care because these decisions determine how fast AI gets deployed, who gets to build the most powerful models, and whether safety actually comes before speed.
What to watch next: The final G7 communiqué—expected later this week—will reveal whether the coalition proposal was endorsed or kicked to a working group. And the fate of Fable 5 and Mythos 5 remains uncertain. If the coalition framework addresses export controls, those models might get restored. But trends suggest that without a unified compromise, the U.S. could continue its unilateral enforcement, and the 'U.S.-led coalition' could become a diplomatic fight disguised as cooperation.