EU AI content labeling rules enter into force with fines up to €15 million
This digest was compiled by AI from multiple sources — links to the originals are below.
The European Union’s AI transparency regulation entered into force on August 2, requiring labels on all AI-generated or altered content. Developers must embed machine-readable markers in images, video, audio and text, and inform users when they are interacting with an artificial intelligence system. Non-compliance can trigger fines of up to €15 million or 3% of global annual turnover for large companies.
Labeling Requirements
Under the EU AI Act, developers of generative systems must embed machine-readable markers in all AI-created or altered images, video, audio and text. Services that interact directly with people must disclose that a user is communicating with artificial intelligence, unless this is obvious from the circumstances. Deepfakes must carry a visible label indicating AI manipulation. AI-generated texts on matters of public importance also require labeling, except when they have undergone human review or editorial oversight by a responsible legal entity.
Exemptions and Transition Period
Law enforcement agencies are granted limited exceptions from the rules on real-time biometric classification and emotion recognition. Artistic, satirical and fictional works are subject to a simplified disclosure procedure. Private individuals using AI solely for personal, non-professional purposes are not covered by the obligations. Systems placed on the market before the new rules took effect have a transition period until December 2 to implement technical marking and automated detection. Content generated before August 2 does not require retroactive labeling, although the Commission recommends voluntary marking where possible.
Penalties
National supervisory authorities can impose fines for non-compliance, with a fixed upper ceiling of €15 million. For large companies, the penalty may reach up to 3% of the previous financial year’s global turnover. When calculating fines for small and medium-sized enterprises, regulators must apply the principle of proportionality.
What's Next
National regulators across the EU are expected to begin enforcement actions in the coming months. It remains unclear how effectively the rules will curb AI-generated disinformation, especially content originating from outside the bloc.
2 sources
EU AI content labeling rules enter into force with fines up to €15 million
