AI Act Article 50: which projects are concerned?
Since August 2, 2026, Article 50 of the EU AI Act imposes transparency obligations on providers and deployers of certain AI systems: chatbots, generated content, deepfakes. Generative systems placed on the market before that date have until December 2, 2026 to mark their content.
What Article 50 says
Article 50 of Regulation (EU) 2024/1689 does not target a risk category but uses of AI. It splits the obligations between the provider, who develops or places the system on the market, and the deployer, who uses it under their own authority.
| Paragraph | Who | Obligation |
|---|---|---|
| § 1 | Provider | Design systems that interact directly with people so that they know they are dealing with an AI, unless this is obvious to a reasonably attentive person. |
| § 2 | Provider, including of general-purpose AI | Mark synthetic content (text, image, audio, video) in a machine-readable format and make it detectable. Exception notably for standard editing assistance. |
| § 3 | Deployer | Inform people exposed to an emotion recognition or biometric categorisation system. |
| § 4 | Deployer | Disclose deepfakes, as well as AI-generated text published to inform the public on matters of public interest, unless it has been human-reviewed under editorial responsibility. |
The information must be clear and given at the latest at the first interaction or exposure (§ 5). A breach can be fined up to €15 million or 3% of worldwide turnover, whichever is higher. For SMEs, the regulation now applies whichever is lower.
The timeline, after the digital omnibus
Regulation (EU) 2026/1744 of July 8, 2026, known as the "Digital Omnibus on AI", in force since July 27, 2026, changed several AI Act deadlines. For Article 50:
- August 2, 2026 : Article 50 applies.
- December 2, 2026 : deadline to comply with § 2 (marking) for generative systems placed on the market before August 2, 2026.
The same text postponed the obligations for high-risk systems to December 2, 2027 (Annex III) and August 2, 2028 (Annex I). A voluntary code of practice on marking and labelling generated content was published in June 2026.
What this touches in a project portfolio
The difficulty is not understanding the article but finding the projects concerned. They are rarely labelled "generative AI" in a portfolio.
The tricky point is often qualifying the role. A company that builds an assistant on a third party's model and puts it into service under its own name may be considered the provider of that system. The qualification is made project by project.
Questions to ask yourself
- Which ongoing or live projects have an AI interact with customers or employees?
- For each, are we the provider, the deployer, or both?
- Which generated content leaves the company, and which is published to inform the public?
- Which generative systems were already in service before August 2, 2026 and must comply with § 2 by December 2, 2026?
- Which teams and vendors carry these projects, and who decides on the measures to take?
See what this hits in your organization
Prévoq links this kind of change to the projects, teams and contracts of your organization, with the level of evidence of each link.
Sources
- Regulation (EU) 2024/1689, Article 50 · ai-act-service-desk.ec.europa.eu · accessed October 15, 2026
- Regulation (EU) 2026/1744 (Digital Omnibus on AI), Official Journal of the EU · eur-lex.europa.eu · July 8, 2026
- European Commission, Q&A on Article 50 transparency obligations · digital-strategy.ec.europa.eu · updated July 24, 2026
- European Commission, code of practice on AI-generated content · digital-strategy.ec.europa.eu · page updated October 12, 2026
This analysis presents the state of the texts and public sources at the date indicated. It does not constitute legal advice.
