TL;DR
Led by Anthropic’s Claude Mythos, US control over the most cyber-capable AI models approaches the July 7-8 NATO summit in Ankara. Washington has torn between export controls and expanded allied access through Project Glasswing, angering its European allies who demand access as they build their own defense artificial intelligence. He will hardly mention it officially at the summit.
Donald Trump comes to next week’s NATO summit in Ankara with unusual leverage as the US decides which allies will have access to the world’s most advanced AI. This was reported by Politico. The alliance is meeting on July 7 and 8 with security issues related to artificial intelligence on the agenda.
The new wave of Anthropic and OpenAI models can find and exploit security flaws better than most human experts. The Anthropic Claude Myth within hours, weaknesses in US classification systems were discovered during a government test.
“AI is fundamentally changing the threat landscape and NATO must adapt,” Estonia’s cyber ambassador Helen Popp told Politico. Every ability available to adversaries is also available to allies, he argued, if they act first.
💜 of EU technology
The latest rumblings from the EU tech scene, a story from our wise founder Boris and some questionable AI art. Free in your inbox every week. Register now!
US agencies including the NSA and CISA are testing Mythos for cyber defense and digital espionage. The European allies clamored for access and EU institutions have clearly requested thisOnly a few countries, including the UK, were initially allowed to conduct assessments.
Anthropic expanded its Project Glasswing program in June About 150 organizations in more than 15 countriesincluding the EU. The confrontation followed weeks of whiplash from Washington.
In early June, the Trump administration imposed export controls on Anthropic’s most cyber-capable models, barring foreign nationals from using them and forcing a worldwide shutdown. Controls Raised on June 30 After 18 days of blackout.
The White House has also limited the deployment of OpenAI’s latest model to a small group of approved US firms, according to Politico. The push and pull frustrated allies, made a rare Five Eyes warning about AI cyber threats, and left. frontier models move faster between governments than regulators can follow.
Quiet corridors, loud subtext
The summit’s agenda includes a track on emerging and disruptive technologies, but one official told Politico that artificial intelligence and cyber will receive only brief mentions in the final statement. Former NATO cyber policy chief Heli Tiirmaa-Klaar said the allies are reluctant to formally discuss issues where there is no consensus, instead predicting that talks will take place on the sidelines.
Politico reports that the US State Department’s cyber bureau is not sending a representative amid internal restructuring. Sen. Jeanne Shaheen said she would attend in part to reassure allies that the United States “will not alienate them” over access to AI models.
Trump also signed NSPM-11, He ordered the US military to adopt AI faster and shield models from China. It hedges by building its own capacity, including in Europe Defense AI alliance between Helsing and Mistral.
The war in Ukraine is now in its fourth year and concretely keeps seals, and allies have pledged 1.5% of GDP to protect critical infrastructure. Laura Galante of the Center for European Policy Analysis called Ukraine a blueprint for an artificial intelligence war.
A State Department spokesman said each ally must adopt “reliable advanced AI capabilities.” What capabilities will be considered reliable and who gives the credit is exactly what Ankara will not discuss.
