G7 Hosts AI CEOs Amid Protests and Criticism

G7 leaders held a working lunch with tech executives, including OpenAI's Sam Altman, Anthropic's Dario Amodei, and Google DeepMind's Demis Hassabis, on June 17, 2026, during the G7 summit in Evian-les-Bains, France. Amodei and Hassabis used the meeting to call for a U.S.-led international coalition to set AI rules and standards, according to CNBC. Just across the border, an estimated 20,000 people protested the summit in Geneva on June 14, including roughly 600 self-described "Black Bloc" militants; the demonstration turned violent, with a car set ablaze and a bank's windows broken amid clashes with riot police using tear gas and water cannon, CBC and PBS reported. A separate counter-summit organized by G7 critics in Geneva devoted significant discussion to AI's societal risks, according to SWI Swissinfo. Truthout, drawing on a Democracy Now! interview, reported that Oxfam International executive director Amitabh Behar called the G7 "a club of the super-rich super-elites," and cited a protest sign reading, "Your enemies don't arrive by boat. They arrive by private jet."

Days before the summit, the U.S. government ordered Anthropic to disable access to its newest models, Fable 5 and Mythos 5, for all foreign nationals, including foreign employees inside the US, citing national-security concerns, Al Jazeera and Reuters reported on June 13, 2026. Anthropic said the government's report likely concerned Fable 5's code-review capabilities and that it disagreed the software should be blocked on that basis, noting rival models have similar capabilities. NPR reported on June 27 that the administration subsequently eased the restriction for one of the two models.

The combination of direct AI-CEO access to heads of state and a same-week export-control action against one of those companies illustrates that policy exposure now cuts both ways: frontier labs can shape emerging international AI standards while simultaneously facing unilateral national-security restrictions on model access. Teams operating internationally should treat export-control and foreign-access risk for frontier models as an active, not hypothetical, compliance consideration.

Key signals to track include whether the proposed U.S.-led AI coalition advances into a formal framework, whether the Fable 5 and Mythos 5 restrictions are further modified, and whether civil-society pressure demonstrated at the Geneva counter-summit translates into concrete G7 policy commitments on AI's economic and environmental impact, which Truthout reports were not substantively addressed at the summit itself.