Anthropic chief executive Dario Amodei pressed G7 leaders to hold a unified line on artificial intelligence, warning them to "resist the temptation to splinter" as governments weigh competing regulatory frameworks. The call drew an unusual show of solidarity from OpenAI chief executive Sam Altman, Amodei's most direct rival in the frontier-model market.

Why Two Rivals Are Reading From the Same Script

The alignment between Amodei and Altman is notable precisely because it is rare. Anthropic and OpenAI compete for the same enterprise contracts, the same research talent, and increasingly the same government relationships. When the heads of the two leading U.S. frontier-AI labs agree in public — and direct that agreement at the world's largest economies — the message carries commercial weight beyond the policy language.

Both companies depend on cross-border data flows, international cloud infrastructure, and the ability to sell into allied markets without running separate compliance regimes in each jurisdiction. A fractured G7 regulatory landscape would raise the cost of doing that business and, more importantly, hand an organizational advantage to state-backed AI programs operating under a single national mandate.

What Splintering Would Actually Mean for the Industry

The practical risk Amodei flagged is not philosophical. If G7 nations adopt divergent AI rules — separate liability standards, conflicting mandatory disclosure requirements, or incompatible safety auditing regimes — companies headquartered in any one of those countries face a choice: build multiple product variants, exit smaller markets, or spend heavily on compliance infrastructure that produces no competitive product advantage.

Smaller AI companies and startups face that calculus more acutely than the large labs, but the large labs set the negotiating terms at summits like the G7, which is part of why Amodei showed up to make the case directly to heads of government.

The Competitive Stakes Behind the Cooperation Argument

International co-operation on AI governance is also, structurally, a market-access argument. Uniform standards written with U.S. and allied-nation input are more likely to reflect the design assumptions of U.S.-headquartered labs than standards written by individual governments reacting to domestic political pressures. The case for multilateralism and the case for preserving competitive position are not the same thing, but at the G7 level they point in the same direction.