The 2 December 2026 Deadline: Article 50(2) Marking of AI-Generated Content
The Digital Omnibus gave one group a short extension under Article 50: providers of generative AI systems that were already on the market before 2 August 2026. Their deadline for machine-readable marking of AI-generated content is 2 December 2026. Here is who that covers, what the obligation requires, and what it means if you only use such tools.
- Article 50(2) of Regulation (EU) 2024/1689: providers of AI systems that generate synthetic audio, image, video or text must mark the outputs in a machine-readable format, detectable as artificially generated or manipulated.
- For systems placed on the market from 2 August 2026, the obligation already applies.
- For systems placed on the market before 2 August 2026, the Digital Omnibus (Regulation (EU) 2026/1744) set the deadline at 2 December 2026.
- It is a provider duty: the company that develops the system or puts it on the market under its own name.
- Companies that only use a generative AI tool are deployers: 50(2) is not their duty, but Article 50(4) deepfake and public-interest text disclosure can be.
What Article 50(2) requires
Providers of AI systems, including general-purpose AI systems, that generate synthetic audio, image, video or text content must ensure that the outputs are marked in a machine-readable format and detectable as artificially generated or manipulated. The technical solutions must be effective, interoperable, robust and reliable as far as this is technically feasible, taking into account the type of content, the costs of implementation and the generally acknowledged state of the art.
The Regulation does not prescribe one technology. Watermarks, metadata, cryptographic provenance signatures, fingerprinting or a combination can all be part of a solution, as long as the result is machine-readable and detectable.
The obligation does not apply where the AI system only performs an assistive function for standard editing, or does not substantially alter the input data provided by the deployer or its semantics. It also does not apply where use is authorised by law to detect, prevent, investigate or prosecute criminal offences.
Who has until 2 December 2026
Only one group: providers of generative AI systems that were placed on the market before 2 August 2026. The Digital Omnibus gave them until 2 December 2026 to bring those existing systems into line. A generative system placed on the market on or after 2 August 2026 had to comply from day one.
If you build a product that generates images, audio, video or text, including a product that wraps a third-party model and puts it on the market under your own name, check whether you count as the provider. If you do, and the product was live before 2 August 2026, this is your deadline.
If you only use generative AI tools
Most SMEs are deployers, not providers: they use ChatGPT, an image generator or an AI writing feature inside another product. Article 50(2) is the provider’s duty, so you do not have to build marking yourself. Your own Article 50 duties are different:
- Deepfakes (Article 50(4)): if you publish AI-generated or manipulated image, audio or video content that resembles real people, objects, places or events and could falsely appear authentic, you must disclose that it is artificially generated or manipulated. For evidently artistic, creative, satirical or fictional work, the disclosure can be limited so it does not hamper the display or enjoyment of the work.
- Public-interest text (Article 50(4)): AI-generated text published to inform the public on matters of public interest must be disclosed as AI-generated, unless it went through human review or editorial control and someone holds editorial responsibility for it.
- Tool choice: ask your generative AI vendors how they meet Article 50(2). If you use their output in client work, their answer becomes part of your own evidence.
A practical plan for the next ten weeks
If you are a provider with a system on the market before 2 August 2026:
- List every product feature that generates audio, image, video or text, including features built on third-party models.
- For each one, record whether it was on the market before 2 August 2026, and who is the provider.
- Choose a marking approach per content type, and document why it is effective and robust as far as technically feasible.
- Test that the marking survives your own export and delivery paths, not only the raw model output.
- Keep a dated record of what you implemented and when, so you can show it to a regulator or a client.
Questions
Does the 2 December 2026 date also delay the chatbot disclosure duty?
No. The extension covers only the Article 50(2) machine-readable marking for generative systems already on the market before 2 August 2026. Article 50(1) chatbot disclosure, 50(3) emotion recognition notices and 50(4) deepfake and public-interest text disclosure have applied since 2 August 2026.
Is a visible "AI-generated" label enough for Article 50(2)?
Article 50(2) asks for marking in a machine-readable format that is detectable, so a visible label alone is unlikely to be enough for a provider. A visible label can still be part of a deployer’s disclosure under Article 50(4). Confirm the approach for your product with qualified counsel if you need certainty.
Not sure which risk tier your AI system is in?
Take the free 5-question Risk Tier Quiz. No email needed, result in under a minute.
Want this tracked automatically, not read about?
The EU AI Act Governance Cockpit is a Notion + n8n workspace that logs your AI systems, runs this exact risk classification, tracks the required actions, and keeps the evidence. One-time payment from €49.
Related guides
Founder of AI Compliance Lab. Specialist in EU AI Act governance workflows and automation. LinkedIn →