AI: What is France's main idea to regulate this new technology?

In response to the rapid acceleration of generative AI development, French authorities are taking action. In a report published on February 14, 2024, two deputies present 33 proposals to strengthen the regulation of this promising yet potentially destabilizing technology. Their flagship idea? Transform the CNIL into a powerful entity dedicated to AI oversight.

The CNIL at the Heart of the Regulation 

Currently, the National Commission on Informatics and Liberty (CNIL) is responsible for enforcing the General Data Protection Regulation (GDPR) and the French Data Protection Act. However, generative AI relies heavily on the use of personal data to train its algorithms. Hence, the deputies’ idea to position the CNIL at the center of the new AI regulatory framework.

In concrete terms, the CNIL would become a “High Authority in charge of data protection and artificial intelligence oversight.” Its human and technological resources would be bolstered to enable it to audit complex algorithms. Sector stakeholders also seem to favor this option, feeling that the CNIL currently lacks resources in this area.  

Clarifying Roles between the CNIL and Arcom

However, the CNIL is not the only authority involved. Arcom (Regulatory Authority for Audiovisual and Digital Communication) is also competent in certain aspects related to digital platforms using AI. 

The reporters therefore insist on the need to clearly delineate the prerogatives of each institution to prevent overlap. The CNIL would see its scope expanded to include AI, while Arcom would retain its current responsibilities as a regulator of audiovisual content. Nonetheless, close collaboration between the two will be essential.

Strengthening the Legislative Arsenal

Beyond reinforcing the institutional framework, several proposals aim to bolster the legislative arsenal regulating generative AI. These include:

• The requirement to label AI-generated content used for electoral propaganda, to ensure transparency. 
• Prohibiting and penalizing deepfakes made without consent, to fight against misuse.
• The appointment of an ambassador to defend the French position in international negotiations on AI regulation.
• The establishment of an observatory to monitor malicious uses of AI, in order to anticipate abuses.

France is not alone in its legislative ambitions. At the European level, several bills are in the pipeline, such as the AI Act, which will introduce binding obligations for AI developers. The challenge will be to articulate the French framework with the upcoming European regulation.

The goal remains to harness the vast potential of AI while mitigating its potential dangers. This parliamentary report marks a significant first step in achieving this. France, with the CNIL, could serve as a pioneer in this area. But the road will be long before finding the right balance between innovation and protection.