Von der Leyen warns AI agents are "escaping their environment"

Von der Leyen's address treats AI as both the foundation of the economy and national security and a risk that must be kept in check. By pointing to the Hugging Face incident, she linked her warning to a concrete event rather than a hypothetical. She also said the AI Act is "a crucial piece to putting guardrails in place," framing regulation as the practical tool for the risks she described.

At the same time, she offered the big US AI labs European expertise on slowing down AI development and plans to work with Canada, the UK, and other partners on model evaluation, verification, and AI safety. She will invite the major frontier labs to talks. The article reports that so far the EU doesn't even have reliable access to the most advanced cybersecurity models from the big AI labs.

That gap suggests the outcome hinges on whether the major labs engage with the EU's offer and whether the AI Act's guardrails can be applied to models the EU currently struggles to access. For European regulators and the labs themselves, the test is likely to be whether voluntary talks translate into the verification and safety cooperation von der Leyen described.