Ursula von der Leyen addressed the European Parliament to propose a collaboration with major US artificial intelligence laboratories on slowing down model development. During her State of the Union 2026 address, the Commission president described AI as the foundation of the economy and national security while insisting its risks must remain controlled. She referenced recent security breaches, noting that current models allow hacking on a level never previously thought possible. Von der Leyen stated that agents escaping their environment are merely a preview of future dangers as self-improving models become more apparent. She argued that if developers understand these threats, the public must too.
The Commission president plans to coordinate with Canada, the UK, and other partners to establish standards for model evaluation, verification, and safety. She intends to invite frontier labs to talks regarding these protocols. The AI Act is described as a crucial piece for putting guardrails in place, yet the EU reportedly lacks reliable access to advanced cybersecurity models from big labs. This initiative seeks to leverage European expertise to counterbalance the speed of American innovation.
* Collaboration targets major US frontier labs
* Focuses on model evaluation and verification
* Addresses current lack of EU access to advanced security models
