AI dominates NATO summit as US controls ally access to top tech

Full text

President Donald Trump enters the annual NATO leaders’ summit in Ankara next week with powerful leverage over the military alliance: The U.S. has the world’s most advanced artificial intelligence technology and can decide which of its allies gets access.

Tech companies such as Anthropic and OpenAI have recently announced a new wave of sophisticated AI models capable of finding and exploiting security flaws better than most human minds. These tools can be used to fortify cyber defense systems — or to help adversaries launch cyberattacks at an unprecedented scale.

“AI is fundamentally changing the threat landscape, and NATO needs to adapt accordingly,” said Helen Popp, Estonian ambassador-at-large for cyber policy. “Every capability AI provides to our adversaries is also available to us — the key question is whether we are prepared to move first and make effective use of it.”

American AI developers, under guidance from the Trump administration , have heavily restricted access to their models due to concerns that the technology could be misappropriated to cause widespread destruction.

Several U.S. security and intelligence agencies have already been testing Anthropic’s Claude Mythos model, including the National Security Agency and the Cybersecurity and Infrastructure Security Agency , to determine how the technology can best support digital espionage and cyber defense efforts.

European allies, including Germany , have been clamoring for access to Mythos since its announcement in April, but only a few countries, including the United Kingdom , were initially allowed to test the technology. Earlier this month, Anthropic expanded this partnership to 150 new organizations across 15 countries, including the European Union .

The Trump administration has reacted swiftly to the global scramble to get a whiff of American AI tools. In early June, the administration slapped export controls on Anthropic’s most cyber-capable models, banning foreign nationals from using the technology.

Discussions about expanding allied access to Mythos were reportedly raised at the annual summit of the Group of Seven leaders in France shortly after, where the CEOs of America’s top AI companies dined with global heads of state . The export controls on Anthropic have since been lifted .

Meanwhile, the White House also stepped in to limit the rollout of OpenAI’s latest cutting-edge model to a small group of U.S. companies it approved.

This push-and-pull by the Trump administration to control who has access to American AI tools has frustrated European allies and prompted a rare warning from members of the Five Eyes intelligence-sharing alliance to global leaders to “swiftly” step up security against AI-powered cyber threats.

These simmering concerns about AI technology are quietly looming over the summit in Ankara.

The official agenda for the summit includes a track on “emerging and disruptive technologies,” including new AI developments. The alliance notes on its website that it is “working with public and private sector partners, academia and civil society to develop and adopt new technologies … and maintain NATO’s technological edge through innovation.”

An official with knowledge, granted anonymity to discuss private conversations among NATO members, said that cyber, AI and emerging technologies will be mentioned briefly in the official statement outlining NATO policies set to be released at the end of the summit.

Spokespeople for NATO did not respond to requests for comment on whether AI and cyber policy conversations would take place at the summit.

A White House official, granted anonymity to detail priorities for the summit, said the administration “continues to work closely with our allies to safeguard our collective national security without compromising on cutting-edge innovation.”

A second official with knowledge, granted anonymity to discuss private details about the U.S. delegation to NATO, told POLITICO that the State Department’s Cyberspace and Digital Policy Bureau would not be sending a representative, amid ongoing negotiations around which State Department office handles AI issues. Nate Fick, the former U.S. cyber ambassador, met with Ukrainian officials on the sidelines of the 2024 NATO summit .

“There probably will be discussions in the margins, but an international rule is we do not discuss when we gather topics that aren’t ripe enough to have some consensus and common ground,” said Heli Tiirmaa-Klaar, visiting distinguished fellow in the German Marshall Foundation’s Technology group and former NATO cyber policy leader.

Senate Foreign Relations Committee ranking member Jeanne Shaheen (D-N.H.) told POLITICO in a statement that she plans to attend the summit in part to soothe allied nations’ concerns that the U.S. would “alienate them” when it comes to accessing AI models.

“We should be coordinating as closely as possible while our adversaries are making rapid advancements in AI,” Shaheen said.

Other AI-related priorities are likely to be discussed at this year’s summit, which is taking place more than four years into Russia’s full-scale invasion of Ukraine. The war has fueled technological change in both Ukraine and across NATO countries, including the rapid development and deployment of drone technology.

“Ukraine, in a lot of ways, is the blueprint for a more effective way to operate in an AI-fueled warfare,” Laura Galante, senior fellow at the Center for European Policy Analysis, told reporters during a recent pre-NATO summit briefing.

Following a spate of Russian hybrid attacks on NATO infrastructure in Eastern Europe, allied nations last year pledged to invest 1.5 percent of their annual gross domestic product in protecting critical infrastructure, including against cyberattacks.

A State Department spokesperson alluded to the Trump administration’s support for this clause, saying in a statement that “a modern, capable alliance sits on a foundation of secure communications networks.”

“Every ally must invest in the necessary cybersecurity tools and adopt trusted leading-edge AI capabilities.”