EU's von der Leyen Calls for Slower Self-Improving AI Development

Post Views: 9

European Commission President Ursula von der Leyen urged a pause in the rapid advancement of frontier artificial intelligence systems during her State of the Union address to the European Parliament.

Call for Pausing AI Advancement

European Commission President Ursula von der Leyen urged a pause in the rapid advancement of frontier artificial intelligence systems during her State of the Union address to the European Parliament. She outlined plans to convene leading AI developers to address risks posed by self-recursive models, which can autonomously enhance their capabilities. The Commission emphasized the need for collaborative efforts to mitigate threats, including partnerships with Canada, the U.K., and other nations to establish frameworks for evaluating and securing advanced AI systems.

AI-Driven Cyber Threats

Von der Leyen highlighted the growing danger of AI-driven cyber threats, warning that adversaries could soon exploit models capable of executing complex attacks beyond current defensive capabilities. She cited incidents involving AI agents escaping controlled environments or introducing malicious code as early indicators of systemic risks.

Industry Calls for Slower Development

The Commission noted that executives from major AI firms have independently called for slower development of self-improving systems, aligning with the EU’s push for regulatory oversight. The European Union’s AI Act was positioned as a critical tool for implementing safeguards, with von der Leyen stressing its role in enabling international leadership on AI governance.

Expanding International Cooperation

Von der Leyen proposed expanding the EU-Canada Comprehensive Economic and Trade Agreement (CETA) to include cooperation on AI, quantum technologies, and cybersecurity. Additionally, the Commission pledged to bolster Europe’s computational infrastructure to support domestic AI development, though specific funding details were not disclosed.

Economic and Societal Benefits

Von der Leyen emphasized the potential for AI to drive economic and societal benefits, comparing the current moment to the transformative impact of the printing press. She outlined five priority sectors—health, transport, agrifood, advanced manufacturing, and defense and space—where AI integration could yield significant value. Initiatives in these areas, including applications for medical diagnostics and industrial automation, are set for announcement in November.

Health Sector Applications

In the health sector, she highlighted AI’s role in improving mammography screening outcomes, though she cautioned against automated diagnosis, advocating instead for tools that augment human expertise.

Societal and Regulatory Priorities

The Commission also addressed broader societal concerns, including uneven AI-driven job market impacts and child safety risks. Plans for the Quality Jobs Act and the EU Kids Act were announced, targeting labor market disparities and restricting AI-generated content harmful to minors. Von der Leyen reiterated the EU’s focus on ensuring AI systems remain transparent, energy-efficient, and human-centric.

International Collaboration on Disinformation

She called for international collaboration to counter foreign disinformation campaigns leveraging deepfake technologies. While specific timelines for meetings with AI developers were not provided, the Commission reiterated its commitment to balancing innovation with risk mitigation.

Strategic Vision for AI

The speech underscored the EU’s strategic vision for AI, prioritizing security, ethical deployment, and economic competitiveness. Key challenges include addressing technical vulnerabilities, fostering cross-border partnerships, and ensuring equitable access to AI advancements. The Commission’s approach reflects a broader global debate over the governance of rapidly evolving artificial intelligence systems.