AI: von der Leyen Warns and Seeks Cooperation with Canada and the UK

Ursula von der Leyen places artificial intelligence at the heart of the economy and security while advocating for safeguards. She warns against agents that could "escape their environment" and more risky self-improving models. She aims to collaborate with Canada, the United Kingdom, and other partners, inviting leading laboratories for discussions. The EU currently lacks reliable access to the most advanced cybersecurity models.

Assertive Safeguards and Limited Access to Models

At this stage, the European Union does not have reliable access to the most advanced cybersecurity models from major AI laboratories. In this context, Ursula von der Leyen asserts that the AI Act is a crucial element for establishing safeguards.

Highlighted Threats: Increased Hacking and Out-of-Context Agents

Referring to an incident at Hugging Face, the President of the European Commission warns that developing AI models could enable hacking at an unprecedented level. She believes that AI agents capable of "escaping their environment" are just a taste of the risks to come, and that the dangers associated with self-improving models are becoming increasingly apparent. According to her, while the designers of these technologies are aware of these threats, the public must be informed as well. She mentions two narratives currently influencing the market, while reminding that AI is a pillar of the economy and national security, but its risks must be managed.

Announced Cooperation with Canada and the United Kingdom

Ursula von der Leyen plans to collaborate with Canada, the United Kingdom, and other partners on the assessment, verification, and security of AI models. She intends to invite leading laboratories for discussions. She also wishes to offer European expertise to major American laboratories to slow down the development of AI. These announcements were made during her State of the Union address for 2026.