EU now requires labels for AI-generated content

The European Union's new AI transparency rules took effect on August 2, 2026, as part of the bloc's AI Act.

According to the European Commission, which adopted guidance on Article 50 of the regulation on July 20, companies must now clearly inform people when they are interacting with AI and label certain types of AI-generated content. The requirements apply to all systems covered by the rules, regardless of when they entered the market.

The rules cover several areas. AI systems that interact directly with people, including chatbots, must disclose that users are communicating with AI rather than a human when this is not already obvious. Synthetic text, images, audio and video must include machine-readable labels that allow their AI-generated origin to be detected automatically. Deepfakes require visible disclosure, while AI-generated text on matters of public interest must be labeled when published without editorial oversight. People must also be informed when emotion recognition or biometric categorization systems are being used.

The requirements also apply beyond the EU. Companies based in the US, UK and elsewhere must comply if their AI products or advertising target users in the European Union. Some content is exempt, including material created by individuals for personal purposes and works that are clearly artistic, satirical or fictional. AI tools already on the market before August 2 have until December 2026 to comply with the machine-readable labeling requirement, while content created and published before the rules took effect does not need to be labeled retroactively.

Violations of the transparency requirements can result in fines of up to €15 million or 3% of a company's annual global revenue, whichever is higher. Smaller businesses and startups face lower maximum penalties.

These are not the AI Act's highest penalties. Companies using prohibited AI practices can face fines of up to €35 million or 7% of global annual revenue.

To help companies comply, the European Commission has introduced a voluntary code of practice for labeling AI-generated content. Around 190 organizations had joined by the end of July. Individual EU countries are also developing additional measures, with Spain preparing its own penalties for unlabeled AI-generated content.