What was actually adopted on 29 June
On 29 June 2026 the Council of the EU gave its final green light to the Digital Omnibus on AI, the package that simplifies and streamlines how the AI Act is implemented. The European Parliament had already endorsed the text on 16 June, so the Council vote was the last step before the regulation is published in the Official Journal and enters into force. For an owner, the headline is simple: the AI Act you read about in 2024 has been amended, and this adopted version is the one your company will be measured against.
Much of the public coverage framed the package as a delay, because some of the heaviest obligations were pushed back. That framing is where a quiet trap sits. The same package that postponed the big-ticket high-risk rules also pulled one specific duty forward, and a business that filed the whole subject under later was not planning for the date that now lands first.
The December 2026 date that moved closer
The Digital Omnibus cut the grace period for marking AI-generated content from six months to three. In plain terms, the obligation to mark content that an AI system produces or manipulates so it can be recognised as artificial now bites on 2 December 2026, not in the second half of 2027 as a six-month reading would have suggested. If your business uses generative tools to produce images, audio, video, or text that reaches customers or the public, this is the date to put in front of whoever owns that workflow.
The same 2 December 2026 date carries a hard prohibition. The package amends the AI Act to ban the use of AI systems to generate or manipulate non-consensual intimate imagery of an identifiable person and to ban AI-generated child sexual abuse material. This is not a paperwork duty with a grace period; it is an outright ban from that day. For most legitimate operators it changes nothing about daily work, but it is the kind of line that any vendor or platform you rely on should already be on the right side of.
What this means for a time-poor owner
The relief most owners heard about is real but later: the stand-alone high-risk obligations under Annex III were deferred to 2 December 2027, and AI embedded in regulated products under Annex I to 2 August 2028. Those dates give breathing room for the harder compliance work. The near-term item is narrower and sooner, namely the December 2026 content-marking duty, and it is the one most likely to slip because it was filed under a heading that said delayed.
A sensible owner does not need to read the regulation to act on it. The practical questions are whether your business publishes anything generated by AI, whether the tools you use already mark their output, and who inside the company would be accountable if a regulator asked. These are points to check with your own legal counsel rather than to settle from a journal article, because the exact scope of marking and any narrow exemptions depend on facts specific to your operation.
Read next: Germany Now Has a Named AI Regulator · Your AI Must Now Identify Itself