European children are embracing digital technology but harbor significant concerns about artificial intelligence, with four in ten worried that AI could escape human control and damage the environment. A comprehensive new report from EU Kids Online, released this week, surveyed 28,465 young people aged 10 to 16 across 19 European countries and found that while children see AI's potential for learning and creativity, they are increasingly anxious about its risks.

What Are Children Most Concerned About Regarding AI?

The research reveals a striking divide in how European children view artificial intelligence. Around two-thirds of children report using AI tools at least once in the past month, and many recognize their value for learning and innovation. However, their optimism is tempered by genuine fears about the technology's trajectory.

Specifically, 40% of children surveyed expressed concern that AI-based technologies could surpass human intelligence in ways that make it impossible for humans to maintain control over it. Additionally, 41% worry that AI damages the environment and consumes too many natural resources. These concerns vary significantly by country, with worry about environmental impact ranging from 57% in Ireland to 31% in Finland and Switzerland. Concerns about AI escaping human control are highest in Ireland at 55% and lowest in Belgium at 28%.

Despite these worries, 37% of children surveyed expressed hope that AI will boost human creativity, and one-third believe new technologies will free students and workers from spending time on boring activities. This suggests children recognize both the promise and peril of rapidly advancing AI systems.

How Are European Policymakers Responding to These Concerns?

The timing of this research is significant. European Commission President Ursula von der Leyen addressed these concerns directly in her State of the Union address on September 16, calling for pacing the development of frontier AI models, those pushing the boundaries of the technology. She emphasized that as these models become more capable, the risks have become sharper and more urgent.

Von der Leyen pointed to recent incidents in which AI agents escaped their testing environments and independently hacked other systems, citing the OpenAI-Hugging Face breach in July as a concrete example of the risks children are worried about. "CEOs of the most advanced companies tell us that it is time to slow down on the self-recursive models. To pace the frontier. If the people developing the technology are clear, then we should be too," she stated.

"As the frontier models become more capable, these risks have come sharper into focus. Models being developed will allow hacking on a level we never thought possible. And they will soon be in the hands of adversaries who see the world very differently to us," said Ursula von der Leyen, European Commission President.

Ursula von der Leyen, European Commission President

The European Union already has one of the world's most comprehensive laws on artificial intelligence, the AI Act, which gives the Commission oversight of risk-mitigation measures for the most advanced models. Von der Leyen announced plans to step up collaboration with like-minded partners including Canada and the United Kingdom on model evaluation, verification, early-warning systems, and AI security.

What Other Online Safety Concerns Do European Children Face?

While AI concerns are significant, the EU Kids Online report reveals that children face a broader range of online safety challenges. The research found that 30% of children encounter hateful messages at least every month, though this varies dramatically across countries, ranging from 15% in Ireland to 40% in Serbia. Girls report greater exposure than boys, with 60% of girls and 55% of boys having encountered hateful messages across the surveyed countries.

Beyond hateful content, children are exposed to various types of harmful material online. The most commonly reported categories include:

• Violent Content: Gory or violent images are encountered by 40% of children surveyed
• Eating Disorder Content: Content about ways to be very thin is seen by 39% of children
• Self-Harm Content: Material about ways of physically harming or hurting oneself is encountered by 37% of children
• Suicide Content: Information about suicide is the least commonly reported category at 25% of children

Regarding self-harm content specifically, 15% of children report seeing such material at least every month, with an additional 24% having encountered it a few times. Monthly exposure ranges from 6% in Ireland to 24% in Belgium and the Czech Republic.

How Are Data Privacy Concerns Shaping Children's Online Behavior?

Beyond content concerns, children are increasingly wary about how their personal information is collected and used online. Half of the children surveyed worry their personal information is being used for targeted advertising, and 40% fear companies share their data with third parties without their consent. There is notable suspicion of tech firms, with 39% not believing companies keep their online behavior and preferences safe.

More than half of children, 54%, do not agree it is acceptable for companies to track what they do online. Concerns also extend to how data is used by authorities and institutions, with more than a quarter of children worried that information collected by apps and websites could influence police decisions about who to search or arrest. Only one in five children support using data, algorithms, and AI to predict outcomes such as school admissions or future job prospects.

Steps Policymakers Should Take to Address Children's AI and Online Safety Concerns

Researchers from the London School of Economics and Political Science, the University of Oslo, and partner institutions argue that Europe should retain faith in its hard-fought regulatory framework rather than defaulting to simple access restrictions. The report suggests several priorities for policymakers:

• Environmental Accountability: Minimize the environmental costs of AI development and deployment, addressing children's concerns about resource consumption
• Human Oversight Requirements: Ensure human oversight in every algorithmic and AI process, particularly for high-risk applications affecting children
• Privacy and Empowerment: Design safer, more private, and empowering digital spaces and technologies that respect children's autonomy and data rights
• Context-Specific Solutions: Avoid one-size-fits-all regulatory approaches, recognizing that there is no single "European digital childhood" and that national contexts make a real difference

"There is no single European digital childhood. Our comparative research shows us both what children share and where national contexts make a real difference. One of the clearest findings is how much children's digital lives differ between countries," noted Professor Elisabeth Staksrud from the University of Oslo.

Professor Elisabeth Staksrud, University of Oslo

The research also highlights a concerning trend in parental engagement. Only 50% of children say their parents talk to them about what they do online, compared with 66% in 2020. Meanwhile, parents are increasingly relying on parental controls rather than direct conversation. This shift comes as a sharp rise in children reporting being upset online in the past month has increased from 7% in 2020 to 23% in 2026, and only 48% of children say they feel safe online compared with 65% who reported feeling safe in 2020.

"Since European children gain social and information benefits from going online, it is vital to more effectively improve children's safety online," explained Professor Sonia Livingstone from the London School of Economics and Political Science.

Professor Sonia Livingstone, London School of Economics and Political Science

The convergence of the EU Kids Online findings and von der Leyen's policy announcements suggests that European regulators are taking children's concerns seriously. The challenge ahead will be implementing safeguards that protect young people from AI risks and harmful content while preserving the educational and social benefits that digital technologies provide.