Wrivio
Get Wrivio
5 min readBy Wrivio Team

The December 2026 EU AI Act Deadlines Explained

Two EU AI Act deadlines fall on 2 December 2026, and if you build or deploy generative AI touching the EU market, one of them probably concerns you. According to commentary on the Act and the AI Omnibus timeline, that date is when older generative AI systems must comply with the rules on marking their output as AI-generated, and when a ban takes effect on AI systems that generate non-consensual intimate imagery or child sexual abuse material. Treat specific dates and scope as something to confirm against the official text, since this is legal timing and secondary summaries vary. This is general information, not legal advice; for your own obligations, get advice from someone qualified.

The marking deadline is the one most ordinary businesses will feel, so it is worth understanding plainly.

What The Output-Marking Rule Requires

The AI Act’s transparency provisions require that AI-generated or manipulated content be marked as such in a machine-readable way, so people and systems can tell synthetic content from human-made. Newer systems have been in scope already; 2 December 2026 is described as the point by which legacy generative systems, those placed on the market earlier, must also comply. We covered the labeling obligation in EU AI Act labeling AI-generated text 2026 and the AI-literacy requirement in EU AI Act AI literacy requirement explained.

The practical question for most teams is narrower than the rule sounds: if you use AI to generate content that reaches the public, you need to understand whether your tooling marks that output and whether you are relying on it to.

Before:

We use AI to draft a lot of our published content, but we edit it, so the marking rule does not apply to us.

After:

We use AI to draft published content. We are checking whether our editing makes it human-authored under the rule or whether the output still needs marking, and we are confirming what our tools actually label. We will get advice on the edge cases rather than guess.

The second version does not assume the convenient reading. It checks.

What The Content Ban Covers

The second 2 December deadline is a ban on AI systems designed to generate non-consensual intimate imagery or child sexual abuse material. For the vast majority of businesses this changes nothing about daily work, because you were never going to build such a system. It is worth knowing it exists, because it marks a shift toward the Act banning specific harmful capabilities outright, not only regulating their use.

What Did Not Change

The AI Omnibus, which entered into force in mid-2026, moved several high-risk deadlines later but, per the commentary, left the general-purpose AI rules and the Article 50 transparency obligations alone. So the marking timeline was not pushed back by the Omnibus. We covered the Omnibus in EU AI Act digital omnibus delay explained and enforcement powers in EU AI Act enforcement powers explained. Do not assume a deadline moved just because other deadlines did.

What To Actually Do

Confirm whether you place generative AI content on the EU market, understand what your tools mark and how, and document the decision you reached and why. The documentation matters as much as the decision, because demonstrating what you concluded is part of what the Act expects. We covered keeping that record in EU AI Act data sovereignty guide.

A Wrivio Context for documenting a compliance decision could say:

Rewrite these notes into a clear internal record of our decision on AI output marking: what we concluded, the reasoning, and what we rely on our tools to do. Keep every date, rule reference, and system name exactly as written. Do not add legal conclusions or obligations not in the notes. Mark anything we flagged for legal review.

Press Ctrl+Shift+Space, paste your notes, and check the diff, then send the edge cases to someone qualified. The authoritative source is the European Commission’s AI Act page at digital-strategy.ec.europa.eu.

Common Questions

What EU AI Act deadlines fall on 2 December 2026?

Per commentary on the Act, that date is when legacy generative AI systems must comply with rules on marking their output as AI-generated, and when a ban on AI systems generating non-consensual intimate imagery or CSAM takes effect. Confirm specifics against the official text.

Does the output-marking rule apply to my business?

If you place AI-generated content on the EU market, likely yes, though editing and authorship can affect scope. Check what your tools mark, do not assume the convenient reading, and get advice on edge cases.

Did the AI Omnibus push the marking deadline back?

According to the commentary, no. The Omnibus moved several high-risk deadlines later but left the general-purpose AI rules and Article 50 transparency obligations, which include output marking, unchanged.

Is this legal advice?

No. This is general information. The AI Act is complex and your obligations depend on your specifics, so confirm against the official text and get advice from someone qualified for your situation.

Download Wrivio for Windows to turn compliance notes into a clear internal record while keeping every date and rule reference exactly as written.