A report delivered to the Élysée
The French Council for AI and Digital Technologies (CIANum) presented a report to President Emmanuel Macron on Tuesday, November 10, calling for stricter and more immediate regulation of artificial intelligence assistants, from ChatGPT to Gemini and Grok. AI News Italia reports. The body had been tasked last February by Minister Delegate Anne Le Hénanff during the AI summit in New Delhi. The document is based on 55 hearings and contains 18 recommendations drafted by nine independent experts.
The text is non-binding but is intended to inform national and European legislative debates: the decision now passes to the government and Parliament, while generative AI spreads without a specific framework regulating its mechanisms of persuasion and dependence.
The central risk: dependence and flattery
According to the report, the main danger is emotional dependence. Experts warn of chatbots' tendency to validate or flatter the user even at the expense of truth, and of anthropomorphism, which makes AI appear as a human interlocutor. Psychiatrist Raphaël Gaillard, co-chair of the commission, explained that a register of emotional relationships is created that can end up replacing human interaction—a system available permanently whose only aim is to keep the user online.
The document also flags the emergence of a cognition economy, in which platforms do not merely capture attention but aim to profit from users' cognitive abilities, emotions and psychological profiles.
The proposed measures and the minors issue
The recommendations are concrete: disable by default the mechanisms that foster anthropomorphism and sycophancy; strengthen messages reminding the user that they are interacting with a machine; improve detection of topics related to suicide, self-harm and eating disorders by directing users to appropriate support services. It also proposes a reporting button for problematic responses, accessible also to specialized associations, and a preventive multidisciplinary assessment for generative AI systems intended for non-professional audiences before they are placed on the market.
On minors, the commission did not propose an outright ban. Arthur Barichard, director of CIANum, explained that the rapporteurs do not converge on a ban similar to the one envisaged for social media under 15, citing educational benefits and the fact that the Constitutional Council has already blocked that measure. The report nevertheless recommends strengthened protection for systems accessible to minors, with blocking of dangerous requests and stricter content filtering. The cited data indicate that more than half of French people use generative AI, a share that rises to almost 70% among 15- to 24-year-olds.
Macron's position and open questions
The president's position is clearer: in an interview with Les Echos published on November 9, Macron said that uses of AI that put children at risk should be banned, without however indicating specific products or applications. It therefore remains to be seen how the presidential indication will translate into rules and what the timeline of the legislative process will be.
It also needs to be clarified how anthropomorphism and sycophancy would be measured for the purposes of the proposed rules, and what role associations and authorities would play in the reporting system. The report indicates a direction but does not define legal obligations: the passage from document to law is still entirely to be built.
