# EU Article 50 Sets AI Disclosure Rules and Exceptions > Source: > Published: 2026-08-03 11:27:15+00:00 # EU Article 50 Sets AI Disclosure Rules and Exceptions EU AI Act Article 50 took effect on August 2, 2026, requiring notices for direct AI interactions, machine-readable marks on many generative outputs, and disclosures for deepfakes and certain public-interest text. Human-reviewed text with editorial responsibility is exempt from the public-interest text disclosure duty, while other exceptions apply to specific obligations rather than creating a blanket exemption. Article 50 of the EU AI Act began applying on **August 2, 2026**, creating separate transparency duties for providers and deployers of certain AI systems. The European Commission's guidance covers direct AI interactions, machine-readable marking of generated or manipulated output, emotion-recognition and biometric-categorisation systems, deepfakes, and AI-generated text on matters of public interest. The rules can apply to providers outside the EU when their system's output is used in the bloc. They also divide responsibility across the AI value chain: providers design systems and output-marking controls, while deployers are responsible for several notices shown to people exposed to the systems or content. ### Different duties have different exceptions Article 50 does not create one general exemption for AI-assisted publishing. Its exceptions and qualifications attach to particular obligations. - • **Standard editing and limited changes:** The provider's machine-readable marking duty does not apply where a system performs an assistive function for standard editing or does not substantially alter the deployer's input data or its meaning. - • **Human-reviewed public-interest text:** A deployer does not have to label AI-generated or manipulated public-interest text when the content has undergone human review or editorial control and a natural or legal person holds editorial responsibility for publication. - • **Artistic or fictional deepfakes:** For evidently artistic, creative, satirical, fictional or analogous works, disclosure must be made in a way that does not hamper display or enjoyment. This qualifies how the disclosure is presented; it does not erase the obligation. - • **Legally authorised uses:** Article 50 provides exceptions for specified law-enforcement uses, with the exact boundary depending on the duty involved. The Commission's July 24 FAQ adds an important operational definition. Human review means deliberate examination of the substance by people with relevant knowledge and professional judgment. Editorial control means practical authority to approve, alter or reject the substance, including fact-checking and source assessment. Superficial spelling or grammar checks are not enough. ### Scope and timing Providers of systems that interact directly with people must make the AI nature of the interaction clear unless it is obvious. Providers of generative systems must generally make synthetic audio, image, video or text output detectable through effective machine-readable marks. Deployers have separate duties for emotion recognition, biometric categorisation, deepfakes and qualifying public-interest text. The Commission says a limited grace period applies only to the Article 50(2) marking-and-detection duty for systems placed on the market before August 2, 2026. Those systems must comply from December 2, 2026. Content generated before August 2 does not have to be labelled retroactively. The AI Act Service Desk warns that its displayed Article 50 text has not yet been updated for Digital Omnibus amendments. For current implementation details, the Commission's guidelines and FAQ are therefore the stronger operational references. For publishers and other deployers, the practical boundary is not simply whether AI touched a draft. Teams need to identify which Article 50 duty applies, document substantive human review and editorial responsibility when relying on that exception, and avoid treating an exception to one duty as an exemption from the rest of the framework. ## Key Points - 1Article 50 applies from August 2, 2026 and assigns distinct transparency duties to AI providers and deployers. - 2Human-reviewed public-interest text with identifiable editorial responsibility is exempt from that text-disclosure duty, while superficial editing does not qualify. - 3Standard-editing, artistic-work and legally authorised-use provisions apply to specific duties and do not create a blanket exemption from Article 50. ## Scoring Rationale Article 50 establishes EU-wide operational requirements affecting generative AI providers, publishers, and teams deploying AI content systems. The human-review and editorial-responsibility boundary is particularly relevant to production workflows, governance tooling, and content provenance practices. ## Sources Primary source and supporting public references used for this report. Practice interview problems based on real data 1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with. [Try 250 free problems](/problems)