Navigating Transparency Requirements under the EU AI Act: A Focus on Article 50

The EU AI Act introduces a comprehensive regulatory framework for artificial intelligence, with a strong emphasis on transparency. Article 50 specifically addresses disclosure obligations for providers and deployers of certain AI systems, aiming to ensure individuals are aware of their interactions with AI and the nature of AI-generated content. This analysis delves into the key provisions of Article 50.

1. Disclosure of AI Interaction: Providers of AI systems intended to interact directly with natural persons must ensure these individuals are informed that they are interacting with an AI system. This obligation applies unless the AI nature of the interaction is readily apparent to a reasonably well-informed, observant, and circumspect individual, considering the context and circumstances.

Exemptions exist for AI systems authorized by law for criminal offense detection, prevention, investigation, or prosecution, subject to safeguards, unless publicly accessible for crime reporting.

2. Marking of Synthetic Content: Providers of AI systems, including general-purpose AI, that generate synthetic audio, image, video, or text content must ensure these outputs are marked in a machine-readable and detectable format. The technical solutions employed must be effective, interoperable, robust, and reliable, considering technological feasibility, content type specificities, implementation costs, and the state of the art.

Exemptions apply to AI systems performing assistive editing functions, those not substantially altering input data, or those authorized for criminal offense-related activities.

3. Transparency for Emotion Recognition and Biometric Categorization: Deployers of emotion recognition or biometric categorization systems are obligated to inform exposed individuals about the system's operation. Data processing must comply with GDPR, Regulation (EU) 2018/1725, and Directive (EU) 2016/680, as applicable.

Law enforcement exemptions, with appropriate safeguards, exist for systems permitted by law for criminal offense detection, prevention, or investigation.

4. Disclosure of Deepfakes: Deployers of AI systems generating or manipulating image, audio, or video content constituting deepfakes must disclose the artificial generation or manipulation.

Exemptions apply to legally authorized uses for criminal offense-related activities. For artistic, creative, satirical, fictional, or analogous works, transparency is limited to disclosing the existence of such content in an appropriate manner that does not impede the work's enjoyment.

5. Transparency for AI-Generated News Content: Deployers of AI systems generating or manipulating text published for public interest news must disclose the artificial generation or manipulation.

Exemptions apply to legally authorized uses for criminal offense-related activities or when the content has undergone human review and editorial control, with a natural or legal person holding editorial responsibility.

6. Modalities of Disclosure: The information required under Article 50 must be provided to concerned individuals clearly and distinguishably at the time of the first interaction or exposure, complying with applicable accessibility requirements.

Article 50 of the EU AI Act establishes a crucial framework for transparency in AI systems. By mandating disclosures regarding AI interactions, synthetic media, emotion recognition, biometric categorization, deepfakes, and AI-generated news content, the lawmakers have made their intent of building user trust loud and clear. Providers and Deployers would have to ensure that they provide adequate and timely disclosures to ensure conformance with these provisions.