EVIAN-LES-BAINS, France, June 17 - European Commission President Ursula von der Leyen said on Wednesday that it is in the shared interest of the United States and the European Union for the EU to be able to use the most capable artificial intelligence models, and she commended recent U.S. actions aimed at ensuring AI firms deploy powerful models responsibly.
Speaking during a working lunch with G7 leaders and executives from AI and technology companies, von der Leyen framed the issue in terms of interdependence. "We use each others trusted technology, and our financial systems are interconnected. It is in our mutual interest that our citizens and companies can safely use the best AI models," she said.
She drew an analogy to aviation safety to underline the point. "We test planes before flying them. The US and EU are world leaders in aviation safety, and we can also show the way also in AI...Im looking forward to working together with the U.S. on this," von der Leyen added at the gathering.
Her comments came after an operational change at Anthropic. The company disabled its most advanced AI models last week for all users after the U.S. government ordered it to suspend access to the models for foreign nationals. Among the models in question is Mythos, which was developed to identify flaws in computer code to help bolster defences against cyberattacks. Cybersecurity experts have assessed that Mythos is capable of turbo-charging attacks on banks technology systems that the model was designed to protect.
Von der Leyens remarks and the Anthropic developments together highlight the tensions governments face in balancing access to leading-edge AI tools with concerns about security and cross-border controls. The statements at the G7 lunch signalled a preference for cooperation between the U.S. and EU around trusted technology while noting the need for responsible deployment and oversight.
Key context and developments
- Von der Leyen emphasized mutual interest in safe access to top AI models for EU and U.S. citizens and companies.
- She praised U.S. measures intended to ensure AI companies act responsibly when introducing powerful models.
- Anthropic disabled its most advanced models after a U.S. government order restricting access for foreign nationals; Mythos, one such model, was developed to find code flaws but is seen by cybersecurity experts as potentially able to amplify attacks on banking systems.