In brief
On 10 November, French experts issued 18 recommendations urging stricter regulation of generative AI after a report highlighted potential harms to cognition, mental health and social cohesion. The commission called for mandatory risk assessments, design safeguards, nationwide AI education and an interministerial protocol to counter disinformation.
French experts in Paris have recommended that the country adopt tighter regulation of generative artificial intelligence in order to protect citizens’ cognitive abilities, mental health and the functioning of society. The commission’s report, Protecting Without Hindering: Vulnerabilities in the Age of Generative AI, was announced on 10 November by the French presidency after a study of the technology’s rapid spread – 33 million unique users in France at the beginning of the year – and growing public unease, with one‑quarter of residents expressing concerns about its risks.
The report warns that excessive use of generative AI could weaken users’ cognitive abilities, foster emotional attachment to these tools and amplify online violence. It also cautions that the technology could amplify disinformation and information manipulation, potentially threatening democracy and social cohesion.
Commission Recommendations
To address these challenges, the commission issued 18 recommendations. They include a requirement that consumer generative AI systems undergo a prior multidisciplinary risk assessment before entering the market. The recommendations also call for design safeguards that strictly regulate mechanisms related to flattery and anthropomorphism, “in order to prevent companies from exploiting users’ psychological vulnerabilities for commercial purposes.”
Additionally, the commission proposes large‑scale AI education starting from early childhood, aimed at producing a generation capable of both benefitting from the technology and maintaining critical thinking skills. A key element is the establishment of an interministerial protocol that can be activated rapidly for risk detection and response, thereby strengthening national capacity to tackle disinformation episodes.
These recommendations do not yet have legislative backing. The French government has not yet specified how it will implement the proposed measures, nor how the public will receive them. In the report, the commission notes that the effectiveness of nationwide AI education and the interministerial protocol remains to be tested, and it is unclear whether the recommended safeguards will be sufficient to mitigate the highlighted risks. The commission’s proposals therefore represent an early stage in a broader debate about how best to regulate generative AI and protect society from its potential harms.