The EU Imposes Strict AI Rules in France

The EU Imposes Strict Rules for AI in France

The European Union remains committed to regulating the digital landscape. Since August 2, new transparency rules have come into effect, affecting both AI system providers and “deployers.” These measures, integrated into the European AI Act, aim to better inform internet users interacting with artificial intelligence.

For AI system providers, Europe requires that models be designed so that users are “explicitly informed every time they interact directly with an AI system.” Additionally, these providers must include a “machine-readable” marker to detect content generated by artificial intelligence.

“Deployers” of AI systems are also obligated to inform users when they interact with tools such as:

• Emotion recognition
• Biometric categorization
• Deepfakes
• Texts on public interest issues whose content is not reviewed by a human

The purpose of these new rules is clear: internet users must be aware when they interact with artificial intelligence or when they see content created by it. According to the European Commission, these measures enable everyone to make better-informed decisions and should also help combat misinformation or misleading content.

Depending on the circumstances, these rules will be enforced by national authorities, the EU's AI Office, or the European Data Protection Supervisor. Penalties for non-compliance can reach up to 15 million euros in fines, or an amount equivalent to 3% of the global annual revenue for companies.

Tech Giants Comply with the Rules

One way to comply with these new EU transparency rules is to adopt the “Code of Good Practices on Transparency of AI-Generated Content” proposed by Brussels. Recently, a list of over 180 organizations that have decided to sign this code was published. Among them are tech and AI giants such as:

• Anthropic
• Google
• Meta
• Microsoft
• OpenAI

Mistral, the European champion of generative AI, is also among the signatories. In an announcement made in July, Google highlighted the measures already taken to align with the EU's transparency goals, including the creation of SynthID, an invisible marker designed to detect AI-generated content, offered to other companies like Apple and OpenAI.

Thus, AI system providers and deployers must comply with these new transparency rules in the European Union, which came into effect on August 2, as part of the AI Act. More than 180 organizations have signed a code of good practices to adhere to these transparency rules.