TL;DR
US control over AI models with increased cyber capabilities, led by Anthropic’s Claude Mythos, looms over the NATO summit in Ankara on July 7-8. Washington has oscillated between export controls and expanding allied access through Project Glasswing, frustrating European allies who demand access while they build their own defense AI. Officially, the summit will barely mention it.
Donald Trump arrives at next week’s NATO summit in Ankara with unusual influence, because the United States decides which allies have access to the world’s most advanced AI. political reports. The alliance meets on July 7 and 8 and AI safety issues are on the agenda.
A new wave of models from Anthropic and OpenAI can find and exploit security flaws better than most human specialists. The Myths of Claude de Anthropic Vulnerabilities discovered in classified US systems in a matter of hours during a government test.
“AI is fundamentally changing the threat landscape and NATO must adapt accordingly,” Estonian cyber ambassador Helen Popp told Politico. All the capabilities available to adversaries are also available to allies, he argued, if they act first.
The 💜 of EU technology
The latest rumors from the EU tech scene, a story from our wise founder Boris and some questionable AI art. It’s free, every week, in your inbox. Register now!
US agencies, including the NSA and CISA, have been testing Mythos for cyber defense and digital espionage. European allies have clamored for access, and The EU institutions have openly demanded itand only a few countries, including the UK, were initially allowed to carry out assessments.
Anthropic expanded its Project Glasswing program in June to around 150 organizations in more than 15 countriesincluding the EU. The fighting came after weeks of whipping by Washington.
In early June, the Trump administration imposed export controls on Anthropic’s most cyber-capable models, banning their use by foreign citizens and forcing a global shutdown. the controls They were lifted on June 30 after an 18-day blackout.
The White House has also limited the release of the latest OpenAI model to a small group of approved US companies, according to Politico. The back-and-forth has frustrated allies, prompted a rare Five Eyes warning about AI cyber threats, and left Border models move between governments faster than regulators. can continue.
Quiet hallways, loud subtext.
The summit agenda includes a topic on emerging and disruptive technologies, but an official told Politico that AI and cyber will only receive brief mentions in the final declaration. Former NATO cyber policy leader Heli Tiirmaa-Klaar said allies avoid formally discussing issues that lack consensus and instead predict sideline talks.
The US State Department’s cyber office will not send any representatives amid an internal reorganization, Politico reports. Sen. Jeanne Shaheen said she will attend in part to assure allies that the United States will not “alienate” them over access to AI models.
Trump has separately signed NSPM-11, order US military to adopt AI faster and models of shields from China. Europe is protecting itself by building its own capacity, including Defense AI alliance between Helsing and Mistral.
The war in Ukraine, now in its fourth year, keeps the stakes concrete, and allies have pledged to allocate 1.5% of GDP to protect critical infrastructure. Laura Galante of the Center for European Policy Analysis called Ukraine the model for operating in an AI-driven war.
A State Department spokesperson said each ally must adopt “state-of-the-art, reliable artificial intelligence capabilities.” What capabilities count as trustworthy and who grants the trust is precisely what Ankara will not fully discuss.
