Article 50 AI Act includes four transparency obligations, each applying to different types of AI systems or their outputs.
| Provision | Type of AI system/output | Transparency obligation | Exceptions or special regimes |
|---|---|---|---|
| Art. 50(1) | AI systems directly interacting with natural persons | Providers must develop and design the AI system in such a way that the natural persons concerned are informed they are interacting with an AI system. | Exceptions: (1) the artificial origin of the interaction is obvious; or (2) the system is authorised by law to detect, prevent, investigate or prosecute criminal offences, unless the system is available to the public to report a criminal offence. |
| Art. 50(2) | AI systems generating or manipulating synthetic image, video, audio or text content | Providers must ensure the AI system’s outputs are marked in a machine-readable format and detectable as artificially generated or manipulated with technical solutions that are effective, interoperable, robust and reliable. | Exceptions: (1) the AI system performs an assistive function for standard editing or does not substantially alter the input data or the semantics thereof; or (2) the AI system is authorised by law to detect, prevent, investigate or prosecute criminal offences. |
| Art. 50(3) | Emotion recognition or biometric categorisation AI systems | Deployers must inform the natural persons exposed to the system of AI system’s operation. | Exception: the AI system is permitted by law to detect, prevent or investigate criminal offences. |
| Art. 50(4) | AI systems generating or manipulating deep fake or text published to inform the public on matters of public interest | Deployers must disclose that the content has been artificially generated or manipulated. | Exceptions: (1) the AI system is authorised by law to detect, prevent, investigate or prosecute criminal offence; or (2) the text publication has undergone human review or editorial control and is subject to editorial responsibility. Special disclosure regime applies to deep fakes that are part of artistic, creative, fictional, satirical or analogous works or programmes. |
Each of these four obligations is examined in Sections 3 to 6 of the se 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.
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 a n AI system generates images as part of a direct interaction with anatural 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.