EU plans new AI safeguards and child online rules

The European Commission plans to step up its response to the risks posed by increasingly capable artificial intelligence systems. In her 16 September State of the Union address, Commission President Ursula von der Leyen said the EU would engage with leading frontier AI laboratories on efforts to slow the development of self-improving models and work with international partners on AI security.

Von der Leyen said future AI models could enable hacking capabilities at unprecedented levels and pointed to concerns over self-improving systems. She announced plans to invite major frontier AI developers to discuss how public authorities could support industry efforts to ‘pace the frontier’. The Commission also intends to work with partners including Canada and the UK on model evaluation, verification, early warning and AI security.

At the same time, the Commission wants Europe to expand its own technological capabilities. Von der Leyen said the EU needs greater computing capacity, particularly to protect national security, and announced plans to develop new ways of using public and private funding to support European start-ups and companies.

Her approach also puts greater emphasis on applying AI across existing European industries rather than focusing only on the development of frontier models. She highlighted industrial data as an asset that could support tailor-made AI systems and pointed to applications in areas including manufacturing, healthcare, agriculture and energy.

Children’s online safety was another area where the speech set out a new legislative initiative. Von der Leyen announced the EU KIDS Act, which the Commission proposed the following day. The planned framework would prohibit social media accounts for children under 13, allow only parent-supervised mini accounts for those aged 13 to under 15, and require safe-by-design protections for users aged 15 to 18.

The proposal is intended to shift greater responsibility to online platforms. Von der Leyen referred to addictive features, excessive scrolling and AI-generated sexualised images involving children, arguing that platforms should demonstrate that their services are safe for minors rather than placing the primary responsibility on children and parents.

The speech also connected digital policy with Europe’s wider security agenda. Von der Leyen cited cyberattacks, sabotage, disinformation and drone incursions as increasingly serious cross-border threats and announced plans for a new European security strategy. She also proposed a European Security Council and an emergency security protocol for major security incidents.

The Commission’s existing cybersecurity work includes its July Action Plan on Cybersecurity and Artificial Intelligence, which addresses both the potential use of advanced AI for attacks and its use to identify threats and protect critical infrastructure.

Technological sovereignty featured across the wider agenda. Von der Leyen linked Europe’s ability to develop and deploy technology with its economic competitiveness and independence, while also pointing to cooperation with Canada on AI, quantum technologies, cybersecurity and economic security. She announced plans for a European corporation on critical raw materials to support access to materials needed for technologies including chips and batteries.

The digital elements of the address therefore combine several strands of EU policy: managing the risks of frontier AI, protecting children online, expanding European computing and industrial capacity, and strengthening cybersecurity and the information environment. Some of the measures announced in the speech, including the EU KIDS Act and further work with frontier AI developers, are intended as new steps, while others build on initiatives already launched by the Commission.