Von der Leyen’s AI Plan: Pace Frontier Models, Protect Kids, Scale Industrial AI

The core message of Ursula von der Leyen’s State of the Union address was selective restraint, not a general slowdown of artificial intelligence. The European Commission president paired a call to manage the risks of the most capable AI systems with plans to protect children online and expand AI use across strategically important industries. 1

1. “Pace the frontier” for the most capable AI models

Von der Leyen said she would invite leading frontier-AI labs to discuss how public authorities could support industry efforts to “pace the frontier.” The announcement concerned talks and safety cooperation, rather than an immediate EU moratorium or a new blanket ban on AI development. 16

Her concern was focused on increasingly capable, including self-improving or self-recursive, models. She warned that such systems could enable hacking on an unprecedented scale and could soon be accessible to adversaries. 115

The proposal arrived amid a wider industry dispute over how rapidly frontier capabilities should advance. Reuters reported that Anthropic chief executive Dario Amodei had argued for a slowdown and received support from OpenAI chief executive Sam Altman and Elon Musk, while the warnings also prompted debate over whether they reflected proportionate precaution or excessive “AI doomerism.” 17

Von der Leyen also called for deeper work with Canada, the United Kingdom and other like-minded partners on AI safety, including the evaluation of frontier models and related security measures. 9

2. An EU Kids Act for age-based online access

The Commission’s proposed EU Kids Act would extend beyond traditional social media. It is intended to cover social-media and video-sharing platforms, online games and AI chatbots, using different safeguards for different age groups. It was announced as a proposal, so it did not itself make new rules immediately enforceable. 1

The approach described by von der Leyen and contemporaneous reporting includes:

• No social-media access for children under 13. 1
• Limited, parent-set-up and parent-supervised accounts for 13- and 14-year-olds, with restricted features and time limits. 1
• A duty for platforms to provide safe-by-design services for users aged 15 to 18.

The design principle is important: responsibility would not rest solely with children and parents. Services would face safety-by-design obligations, while reporting on the proposal said companies would be expected to verify users’ ages and address addictive product features such as infinite scrolling. 7

3. Faster AI deployment in Europe’s strategic sectors

Alongside frontier-risk controls, von der Leyen presented AI as a foundational technology for Europe’s economy and security. She pointed to applications including autonomous driving and defence, while arguing that Europe should extract practical value from the technology. 1

The Commission identified five high-value areas for industrial AI where it says Europe has relevant data and capability:

1. Health
2. Transport
3. Agri-food
4. Advanced manufacturing
5. Defence and space 1

It said it would announce “game-changing initiatives” across those sectors in November. 1

What the agenda means

The policy direction is best understood as a split strategy. At the frontier, the Commission wants stronger evaluation, international coordination and a slower, safety-conscious approach to the highest-risk capability gains. In the broader economy, it wants faster application of AI in sectors tied to public services, industrial competitiveness and security. 19

That distinction matters: the announcement was not a call to slow every AI product or use case. It was a push to place greater caution around frontier models while accelerating beneficial, industry-focused deployment in Europe. 1