EU implements new rules to regulate artificial intelligence

New European Union regulations governing artificial intelligence have taken effect, compelling companies to implement measures for assessing and mitigating risks tied to AI models. These rules demand transparency, ensuring that content created by AI is clearly identified and that "deepfakes", manipulated media simulating false events or statements, are labeled with specific icons or notices.

Models already on the market have until December 2 to comply with these new requirements. The newly established AI Office within the European Commission is empowered to conduct thorough evaluations of large AI models. This office can enforce compliance, impose restrictions on the deployment of new AI models, and even require providers to grant access to their systems before commercialization in certain cases.

Penalties for non-compliance are substantial. Fines for prohibited activities could reach up to 7% of a company's global annual turnover or 35 million euros (approximately $40 million). Other infractions may result in fines of up to 3% of global annual turnover or 15 million euros. Companies obstructing investigations also face sanctions.

The regulation is set to expand further. Starting in December, AI systems capable of generating non-consensual nude images will be prohibited. Additional restrictions targeting high-risk AI applications in sensitive sectors like health and security are slated for implementation between 2027 and 2028, signaling a comprehensive approach to AI governance.

“

From now on, we will be able to carry out exhaustive evaluations and we will do so whenever necessary.