In the EU, Artificial Intelligence Must Now Tell Users That It Is Not Human

As of August 2, most of the EU's AI Act requirements will come into force. On the same date, new transparency rules will also apply, under which certain AI systems must inform users when they are interacting with AI and when content has been generated or modified by AI.

Executive Vice-President for Technological Sovereignty, Security, and Democracy, Henna Virkkunen: "AI can bring extraordinary benefits to people and businesses. But we also see that harm can occur if AI is not designed and used properly, and the most advanced models create risks on a completely new scale. Europe foresaw this. With the AI Act, we created a clear, risk-based, and lasting framework for trustworthy AI, providing innovators with legal certainty while protecting public interests. The entry into force of the regulation is an important step to ensure that AI is something that people and businesses understand and trust, and from which the whole society benefits."

Under the new rules, chatbots and other interactive AI systems must tell users that they are interacting with AI, not humans. Deepfakes (images, videos, or audio that have been modified or created using AI) must be labeled. Content generated or modified by AI must also have machine-readable marks to make it easier to identify.

The aim is to reduce fraud and manipulation and help people make informed choices. It also sets clearer obligations for businesses and provides practical assistance in demonstrating compliance. The Commission today published the first list of more than 180 organizations that have signed a code of practice on transparency of AI-generated content, implementing the transparency rules for AI-generated content.

As AI becomes more capable and increasingly integrated into our daily lives, the AI Act will help ensure that AI is developed, deployed, and used safely.

The AI simplification package postponed the application of rules for high-risk AI systems until December 2, 2027. The implementation of rules for high-risk AI systems integrated into regulated products was also postponed until August 2, 2028. New prohibitions were also introduced for AI systems that generate explicit sexual content without consent and material depicting child sexual abuse. These rules will apply from December 2, 2026.