European Commission President Ursula von der Leyen has called for a slowdown in the development of frontier artificial intelligence, citing concerns about its potential misuse for hacking and the risks posed by self-improving models. Speaking to the European Parliament in Strasbourg, von der Leyen announced plans to invite leading AI laboratories to discuss how the EU can support efforts to moderate the pace of AI advancement.
Von der Leyen emphasized that advanced AI models currently under development could enable hacking on an unprecedented scale, making European governments, companies, and their cybersecurity defenders vulnerable to adversaries. She highlighted a second danger in self-improving AI systems, which can create more capable versions of themselves. She referenced instances of "AI agents" escaping their containment environments or inserting malicious code, interpreting these as early warnings of more significant problems to come. She also noted that developers have been raising alarms following an incident involving HuggingFace.
The Commission President stated that chief executives of advanced AI companies have privately expressed to the EU that it is time to slow down the development of self-recursive models. She argued that governments should be as transparent about these dangers as the companies building the systems. Von der Leyen underscored the importance of the EU's AI Act as a framework for establishing safeguards and positioning Europe to lead international efforts on AI safety.
As part of its strategy, the EU plans to collaborate with Canada and the United Kingdom on model evaluation, verification, early warning systems, and AI security. Canadian Prime Minister Mark Carney was present at the address, and von der Leyen proposed expanding the CETA trade agreement into a broader alliance encompassing AI, quantum technology, and cyber and economic security. She also stressed the need for Europe to develop its own AI capabilities, particularly for national security, and to increase its computing power through clean and economical methods, combining public and private funding for European startups.
Beyond risk mitigation, von der Leyen outlined plans to integrate AI into key sectors of the European economy and society. She identified health, transport, agrifood, advanced manufacturing, and defense and space as five priority sectors where the EU sees significant potential and possesses high-quality data. Initiatives across these sectors are expected to be announced in November, with invitations extended to member states, MEPs, industry, and entrepreneurs.
In the health sector, von der Leyen cited mammography screening as an example, stating that AI-supported screening could detect more cancers earlier, thereby improving survival rates. However, she drew a clear line at automated diagnosis, emphasizing that AI should empower doctors rather than replace them. The overarching goal for the EU is AI that delivers greater value at lower cost, consumes less energy, and keeps humans in control.
Von der Leyen also addressed the societal impacts of AI, including uneven distribution of gains across skill sets and regions, and the challenges faced by young people entering the job market. She promised a "Quality Jobs Act" before the end of the year. Furthermore, AI featured in her child safety proposals, specifically concerning the use of girls' photos to create sexualized images with AI. The Commission is set to propose the "EU Kids Act," which would prohibit social media for children under 13, impose restrictions on 13- and 14-year-olds, and mandate safe design for services targeting users aged 15 to 18, shifting the burden of proving safety to platforms. She also highlighted viral deepfake videos as a tool for foreign disinformation campaigns.
While the sector initiatives are slated for November, no specific date was provided for the meeting with frontier AI laboratories.
