Anthropic and OpenAI have been formally invited to participate in NATO’s annual summit in The Hague this week, marking the first time leading American frontier-AI labs take seats alongside heads of state and defense ministers at the alliance’s flagship gathering. The joint invitation, confirmed Sunday by NATO officials, signals an unmistakable pivot in how the alliance intends to frame the AI race: not as a U.S.-China bilateral contest to be adjudicated in Washington, but as a transatlantic industrial and security priority that Europe intends to shape on its own terms.
The move comes after months of friction in Brussels and Berlin over what European officials describe as an “asymmetric dependency” on a handful of American AI providers. With Anthropic’s Claude and OpenAI’s GPT models now embedded in everything from French military logistics software to German Bundeswehr procurement pipelines, the political class across the bloc has grown increasingly vocal that the current arrangement treats Europe as a permanent customer rather than a co-author of the technology underlying its own defense.
A formal seat at the table for frontier AI
Two of the three largest U.S. frontier model developers will join a NATO Innovation Hub roundtable on Monday, alongside delegations from the European Defence Agency, the U.K.’s AI Security Institute, and Japan’s METI. Sources briefed on the agenda say the session will focus on three priorities: shared evaluation standards for military-grade AI systems, a coordinated export-control regime for high-end AI chips, and joint funding for open-weight models that allied governments can audit and host inside their own sovereign clouds.
“We’ve been told for two years to wait, that the frontier is a few quarters away, that we’ll be included in the next round,” said a senior European Commission official who was not authorized to speak on the record. “Our ministers have decided that waiting is no longer the strategy. We are the second-largest economic bloc in the world. We will build, we will procure, and we will regulate, on our own timetable.”
The shift has been building for months. France’s Mistral, Germany’s Aleph Alpha, and the U.K.’s new “BritGPT” consortium have all secured nine-figure sovereign-cloud contracts in 2026. But European defense planners insist they need a credible counterweight at the frontier, not merely a regional champion. Hence the unprecedented move: inviting Anthropic and OpenAI not as vendors pitching contracts, but as strategic participants in a multilateral framework.
What the labs are bringing to the summit
- Anthropic is expected to brief ministers on its recently expanded Constitutional AI safety stack, including red-team evaluation tools designed specifically for autonomous-weapons-adjacent systems.
- OpenAI plans to present its “Sovereign Deployment Reference Architecture,” a packaged build of GPT-class models that can run inside NATO-member air-gapped enclaves without phoning home to U.S. data centers.
- Both companies will sign a non-binding commitment to share model capability disclosures with allied safety institutes at least 30 days before public release, mirroring export-control transparency norms in semiconductors.
The economic stakes behind the handshake
Defense AI procurement across NATO’s 32 members is on track to clear $41 billion this fiscal year, according to alliance budget documents reviewed by allied analysts. Of that, roughly $27 billion is earmarked for software and services, with the remainder going to AI-enabled hardware such as autonomous drones, space-based sensors, and command-and-control platforms. European capitals have grown frustrated that an estimated 70 percent of that software spend is currently flowing to U.S. providers, with little visibility into training data, evaluation regimes, or downstream audit rights.
“Europe is not looking to copy American frontier models wholesale,” said one German defense ministry adviser. “Europe is looking to ensure that when a French colonel is making a targeting decision with AI assistance, the system in front of him was built with European legal and ethical defaults baked in from the start.”
Risks of the new alignment
Not everyone in Washington is celebrating. Several senior voices in the U.S. intelligence community have warned that deeper entanglement between frontier AI labs and NATO could accelerate an export-control arms race with China and Russia, and could force labs into disclosure schedules that conflict with competitive-release timelines. A few congressional aides have privately floated the idea of new CFIUS-style review authority over AI deals with allied governments, echoing the 2018-era debate over TikTok’s Hong Kong ownership.
Industry response so far has been quietly supportive. Anthropic’s policy lead called the summit “the most consequential multilateral AI conversation in the alliance’s history,” while OpenAI’s head of global affairs described it as “an overdue acknowledgment that frontier AI is now critical national infrastructure.” Both statements stopped short of any commitment to offshore model weights to allied jurisdictions.
The road to a transatlantic AI pact
Officials expect a draft “Transatlantic AI Cooperation Charter” to circulate among delegations before the summit closes. The draft, still under negotiation, envisions a joint evaluation center based in The Hague, mutual recognition of safety certifications across member states, and a co-funded $4 billion AI-for-defense venture capital pool targeting European dual-use startups.
Whether the charter lands as a binding treaty or a softer political declaration will likely depend on how firmly the incoming Dutch summit chair pushes for consensus. For the first time in the alliance’s history, the loudest voices in the room may not be in English.
Anthropic and OpenAI’s presence at NATO is more than a photo opportunity. It is the public marker of a strategic realignment that has been underway for at least a year, and the clearest signal yet that Europe intends to treat frontier AI as sovereign infrastructure rather than a service it buys from across the Atlantic. The Hague summit will show whether the rhetoric outlives the handshake.