Two sets of obligations under the EU AI Act (Regulation (EU) 2024/1689) take effect on 2 December 2026, both fixed by the Digital Omnibus on AI, Regulation (EU) 2026/1744. Providers of generative AI systems that were already on the EU market before 2 August 2026 must have implemented the Article 50(2) marking obligation by that date. Separately, new prohibitions in Article 5 on AI systems that generate non-consensual intimate material or child sexual abuse material begin to apply. Both dates come from the text of the regulation as published in the Official Journal on 24 July 2026.
What the regulation says
Regulation (EU) 2026/1744 was adopted on 8 July 2026 and published in the Official Journal of the European Union on 24 July 2026 (L series, 2026/1744). Under its Article 4 it entered into force on the third day after publication, which the European Commission’s announcement that the AI Omnibus has entered into force dates as 27 July 2026. We covered the headline deadline shifts in our earlier post, EU AI Act Digital Omnibus: what the deadline delay means. This follow-up isolates the two obligations that land on 2 December 2026, about three weeks away.
1. Article 50(2): marking of AI-generated content for systems already on the market
Article 50(2) of the AI Act requires providers of AI systems, including general-purpose AI systems, that generate synthetic audio, image, video or text content to ensure the outputs are marked in a machine-readable format and detectable as artificially generated or manipulated. The Omnibus adds a new paragraph 4 to Article 111, which reads in the Official Journal text: providers of such systems “that have been placed on the market before 2 August 2026 shall take the necessary steps in order to comply with Article 50(2) by 2 December 2026.”
The recitals of the regulation describe this as a transitional period of four months for providers who have already placed their systems on the market. What the amending text does not do is extend the date for systems first placed on the market on or after 2 August 2026, and the new paragraph refers only to Article 50(2). The Omnibus text also replaces Article 50(7) so that the Commission encourages and facilitates codes of practice on the detection, marking and labelling of artificially generated content, rather than approving them by implementing act.
2. Article 5: two new prohibited practices
The Omnibus inserts points (ba) and (bb) into Article 5(1), first subparagraph, and adds paragraphs 1a and 1b. Article 113 is amended so that these provisions apply from 2 December 2026, whereas the rest of Chapters I and II applied from 2 February 2025. In summary, as set out in the regulation:
- Point (ba) prohibits placing on the market, putting into service or using an AI system that generates or manipulates realistic images, videos, audio or similar material of an identifiable natural person’s intimate parts, or of an identifiable person engaged in sexually explicit activities, without that person’s freely given, specific, informed, unambiguous and explicit consent.
- Point (bb) prohibits placing on the market, putting into service or using an AI system that generates or manipulates material or performance within the meaning of Article 2, points (c) and (e), of Directive 2011/93/EU (child sexual abuse material), except where a “without right” defence applies under national law.
- Paragraph 1a limits the provider-side prohibition to systems whose intended purpose is such generation or manipulation, or where it is a reasonably foreseeable and reproducible outcome of the system’s design, training, architecture, capabilities or user-facing functionalities and the system lacks reasonable and adequate technical safeguards to prevent it. For deployers, the prohibition applies where the deployer uses the system for the purpose of generating or manipulating such material.
- Paragraph 1b provides that, for point (ba), manipulation that does not increase the exposure of any depicted intimate parts or alter the nature of any depicted sexually explicit activities is not “manipulation”.
The recitals state that the prohibition is not meant to stop providers from developing the technical capabilities of AI systems to generate or manipulate images, videos or audio generally.
What stays the same on that date
The Commission’s announcement and the regulation’s text place the deferrals elsewhere: Annex III high-risk obligations apply from 2 December 2027 and Annex I high-risk obligations from 2 August 2028. Those later dates should not be confused with 2 December 2026, which concerns only the Article 50(2) transitional period and the new Article 5 provisions described above.
What it means for researchers, universities and research offices
The following is our editorial reading of the regulation text, not legal advice, and no source cited here addresses research institutions specifically.
- Institutions that build or release generative models. A university group or institute that develops a generative AI system and makes it available in the EU is potentially a “provider” for Article 50(2). Whether the Article 2(6) research exemption applies depends on whether the system was developed and put into service solely for scientific research; a system released more widely warrants a documented assessment. Systems already released before 2 August 2026 fall within the December transitional date; later releases do not.
- Institutions that only use third-party tools. The Article 50(2) duty falls on providers, so most research offices will be affected through vendor contracts and procurement questions (does the tool mark its outputs, and from when?) rather than by marking anything themselves. Deployer duties under Article 50 are a separate matter and are not the subject of the new transitional paragraph.
- Policy and conduct. The new Article 5 prohibitions are relevant to institutional acceptable-use policies, student and staff conduct codes, and the vetting of image and video generation tools used in teaching or communications. Research involving sexually explicit or abusive imagery has legal and ethics-review implications beyond the AI Act and should go through ethics and legal review before any work starts.
- Disclosure practice. Machine-readable marking of outputs interacts with journal and funder rules on disclosing generative AI use; see our explainer on what the EU AI Act means for research administration and the industry reaction to the Omnibus.
What we could not confirm
We have not confirmed how national authorities will supervise or enforce either provision, nor the status of any codes of practice or harmonised standards on marking, so this post makes no claim about enforcement activity from November 2026. Check the Commission’s AI Act pages for current guidance.
Sources
- Regulation (EU) 2026/1744 (Digital Omnibus on AI), Official Journal of the European Union, L series, 24.7.2026, EUR-Lex (read 12 November 2026).
- European Commission, Digital Strategy: “AI Omnibus enters into force”, 27 July 2026 (read 12 November 2026).
