The European Union has begun implementing new rules requiring users of generative artificial intelligence systems to disclose certain content created or modified using the technology, particularly deepfake videos and AI-generated text concerning matters of public interest. The measures form part of the transparency requirements established under the EU Artificial Intelligence Act and aim to enable the public to distinguish between authentic and artificially generated or manipulated material.
The new requirements cover AI-generated or modified images, audio recordings and videos that imitate real people or events. They also apply to AI-generated text published on matters of public interest when the material has not undergone human review or editorial oversight.
Symbols to Identify Artificially Generated Content
The European Union has also developed a set of symbols that content creators, publishers and users of AI systems can employ to identify artificially generated or modified material. The symbols include classifications for content created entirely using AI, as well as content that has been partially modified through the use of the technology. A general-purpose symbol can also be combined with a textual description to provide additional information about the nature of the content.
However, the disclosure requirements do not apply to every piece of content produced with the assistance of AI. Under specific circumstances, the rules exempt artistic, creative, satirical and fictional works. They also do not apply when deepfake technology or AI-generated text is used for legitimate purposes related to detecting, preventing or investigating crime.
For AI-generated text published on matters of public interest, disclosure is not required when the material has undergone human review or editorial oversight and a person or organisation assumes responsibility for its publication.
Exceptions and Disclosure Requirements
European Commission Executive Vice-President Henna Virkkunen said that artificial intelligence offers significant benefits to businesses and individuals, but warned that its advanced capabilities could also create new risks if AI systems are not properly designed and used.
The implementation of the rules coincides with the signing by more than 180 organisations of a voluntary Code of Practice launched by the European Commission in June 2026. The code aims to establish common practices for labelling content generated or modified using artificial intelligence.
The new requirements initially apply to newly introduced or newly deployed generative AI systems, while systems already in operation have until 2 December 2026 to comply with the rules.
