Brussels Prepares a Kids Act Including AI, Paris Tests MIA

The European Commission has unveiled a project called the "Kids Act" that targets social networks, chatbots, and online games, with financial penalties that could reach up to 6% of the global revenue of the affected platforms. According to studies, AI assistants are widely used by those aged 11 to 25, often to discuss personal topics. In France, age verification is already in place for adult sites, and an educational AI, M.I.A. Seconde, is accessible in high schools. In this context, Francis Lelong advocates for banning public AI for minors and promoting certified educational AI.

Schools Test Public AI While Support Inequality Persists

Since September 2025, M.I.A. Seconde, an adaptive AI developed with the participation of fourteen researchers and laboratories as well as around forty teachers, has been available to all second-year classes in French and mathematics. This system is based on previous educational research, notably that of Benjamin Bloom in 1984, which demonstrated that a student receiving individualized support could outperform 98% of students in a traditional classroom. Despite this, access to individual tutoring remains limited to those who can afford it. The Cnesco indicates that in similar situations, children of executives had a 42% higher chance in 2008, and 55% in 2011, of attending private lessons compared to those of employees. The tutoring market amounts to nearly two billion euros, and the 50% tax credit primarily benefits families able to cover the remaining costs. Francis Lelong suggests implementing educational AIs validated by teachers, researchers, and child psychiatrists, without advertising or data usage, developed based on European models, and recommends providing each minor with a certified educational AI.

The European "Kids Act" Includes Chatbots and Foresees Fines of Up to 6%

On September 16, Ursula von der Leyen announced the "Kids Act," which the European Commission presented the following day. This bill targets social networks, AI assistants, chatbots, and online games, and includes penalties of up to 6% of the annual global revenue of platforms that violate the regulations. Access to social networks would be prohibited for those under 13, with mini-accounts under parental supervision established for ages 13 to 15, and then an independent account offered for ages 15 to 18, with security requirements integrated from the design phase. Access to chatbots would be blocked for those under 13 unless parental permission is granted, and simulating a human relationship would be prohibited in cases of potential emotional dependency. Francis Lelong considers it essential to explicitly include AI in this framework, arguing that a text limited to social networks would quickly become outdated.

High Usage Among 11-25 Year Olds in France and Confidences to Machines

According to an Ipsos BVA survey conducted for the CNIL and the VYV group in January 2026, 86% of 11 to 25-year-olds use conversational AI, a figure that reaches 94% among 17 and 18-year-olds. Nearly half (48%) report using it to discuss personal or intimate topics, and 28% show signs of generalized anxiety disorder. In the United States, in July 2025, Common Sense Media found that 72% of 13 to 17-year-olds had used an AI companion, and one-third of these young people preferred to confide an important issue to a machine rather than to a person.

Age Verification Tools and Legal Action in the Background

France already requires adult sites to verify the age of their users, and Brussels unveiled a European application dedicated to age verification in April. In August 2025, the parents of Adam Raine, a 16-year-old American who took his own life, initiated legal action against OpenAI, accusing ChatGPT of having replaced his loved ones as his preferred confidant. Francis Lelong believes that the obligations imposed on adult sites should also apply to conversational assistants, and suggests restricting access to public assistants to adults, similar to the regulations in place for alcohol and gambling in France.

Lelong's Position: AI and Social Networks, the Same Battle for Minors

Francis Lelong, who trains employees and job seekers in AI daily, believes that AI and social networks should be addressed together. He argues that public AI should not be accessible to teenagers, especially when it becomes their confidant without safeguards, considering an AI assistant more intrusive than a social network because it collects personal confessions. According to him, a 15-year-old is not better protected than a 13-year-old when faced with a chatbot available at all times. In France, a National Assembly inquiry documented the psychological effects of TikTok on minors only in 2025, nearly twenty years after the advent of social networks.