Guidelines on the implementation of the transparency obligations for certain AI systems under Article 50 of the AI Act — C(2026) 5054 final, Annex

(6) Article 50 AI Act includes four transparency obligations, each applying to different types of AI systems or their outputs. Provision Type of AI Transparency obligation Exceptions or special regimes system/output Art. 50(1) AI systems Providers must develop and Exceptions 1) if the artificial origin directly design the AI system in of the interaction is obvious, or 2) interacting with such a way that the natural the system is authorised by law to natural persons persons concerned are detect, prevent, investigate or informed they are prosecute criminal offences, unless interacting with an AI the system is available to the public system. to report a criminal offence. Art. 50(2) AI systems Providers must ensure the Exceptions if 1) the AI system generating or AI system’s outputs are performs an assistive function for

(1) Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act) (OJ L, 2024/1689, 12.7.2024).

(2) Article 1 AI Act.

manipulating marked in a machine- standard editing or does not synthetic image, readable format and substantially alter the input data or video, audio or detectable as artificially the semantics thereof, or 2) the AI text content generated or manipulated system is authorised by law to with technical solutions that detect, prevent, investigate or are effective, interoperable, prosecute criminal offences. robust and reliable. Art. 50(3) Emotion Deployers must inform the Exception if the AI system is recognition or natural persons exposed to permitted by law to detect, prevent biometric the system of AI system’s or investigate criminal offences. categorisation operation. AI systems Art. 50(4) AI systems Deployers must disclose Exceptions if 1) the AI system is generating or that the content has been authorised by law to detect, prevent, manipulating artificially generated or investigate or prosecute criminal deep fake or text manipulated. offence, or 2) if the text publication published to has undergone human review or inform the editorial control and is subject to public on editorial responsibility. matters of public interest Special disclosure regime applies to deep fakes part of artistic, creative, fictional, satirical or analogous works or programmes.

(7) Each of these four obligations is examined in Sections 3 to 6 of these Guidelines. In addition, Article 50(5) AI Act mandates that the information referred to in its paragraphs 1 to 4 shall be provided to the natural persons concerned in a clear and distinguishable manner at the latest at the time of the first interaction or exposure and conform to the applicable accessibility requirements. Section 7 clarifies these requirements.

(8) The various transparency obligations laid down in Article 50 AI Act may apply cumulatively to (the output of) a single AI system, engaging possibly the responsibility of different actors (providers or deployers). For example, if an AI system generates images as part of a direct interaction with a natural person, then the provider of the system may have to comply with both Article 50(1) and (2) AI Act. Where that system is also able to generate images constituting deep fakes or text published on matters of public interest, the deployer may additionally have to comply with Article 50(4) AI Act.