AI Act Omnibus Finally Adopted: New Deadlines, New Bans – and What Stays Unchanged
The "AI Omnibus" package is final: Parliament approved it on June 16, Council gave final approval on June 29, 2026. The key message stands: Art. 50 applies unchanged from August 2, 2026 – only the high-risk obligations were postponed.
Christian Lechner
On May 7, 2026, the European Parliament and the Council of the EU reached a political agreement on the so-called "AI Act Omnibus" package. These changes affect key deadlines and obligations of the EU AI Act and represent the most significant regulatory development since the AI Act was adopted.
Update (July 2026): The package has since been finally adopted: the European Parliament approved it on June 16, 2026, and the Council gave its final approval on June 29, 2026. The changes described here are therefore no longer a provisional agreement but enacted law.
In this article, we explain what specifically changes, what stays the same – and what it means for website operators.
Key Changes at a Glance
⏰ New Deadlines
| Obligation | Old Deadline | New Deadline |
|---|---|---|
| Art. 50 Transparency (general) | August 2, 2026 | August 2, 2026 (unchanged!) |
| Machine-readable marking, Art. 50(2) (existing systems only) | August 2, 2026 | December 2, 2026 |
| High-Risk AI (Annex III) | August 2, 2026 | December 2, 2027 |
| High-Risk AI (Annex I, Products) | August 2, 2026 | August 2, 2028 |
🚫 New Ban: "Nudifier" Apps
The Omnibus package introduces a new prohibition: AI systems that generate non-consensual sexually explicit content or CSAM (Child Sexual Abuse Material) are banned from December 2, 2026. This is a direct response to the growing spread of so-called "nudifier" apps.
What Was Postponed – and What Was NOT
The core message of the finally adopted version can be summed up in one sentence: only the high-risk obligations were postponed – the transparency obligations under Article 50 apply unchanged from August 2, 2026.
Postponed:
- High-risk AI under Annex III: new date December 2, 2027
- High-risk AI under Annex I (regulated products): new date August 2, 2028
NOT postponed:
- Article 50 (transparency & labeling): August 2, 2026 – chatbot disclosure, deepfake labeling, labeling of AI-generated content
The only nuance: for the machine-readable marking under Art. 50(2) (a provider obligation), there is a transition period until December 2, 2026 – but only for systems placed on the market before August 2, 2026. New systems must be fully compliant immediately from August 2, 2026.
In addition, the final version confirms the new Article 5 prohibition (see above) and softens the wording of the AI literacy obligation (Art. 4) from "ensure" to "support and promote" – still without any direct fines attached.
What This Means for Website Operators
The Good News
For existing AI systems placed on the market before August 2, 2026, there is a transition period until December 2, 2026 for the machine-readable marking under Art. 50(2). This gives providers of such existing systems an additional 4 months for the technical implementation.
The Important News
General transparency obligations under Art. 50 remain at August 2, 2026! This includes:
- Chatbot disclosure: Users must know they are interacting with AI
- Deepfake disclosure: Manipulated content must be disclosed
- Emotion recognition: Users must be informed
For website operators publishing AI-generated content, the core obligation is unchanged: From August 2026, your users must be able to identify which content is AI-generated.
New Systems: Immediately Compliant
Important: The transition period only applies to existing systems. Any AI system placed on the market from August 2, 2026 must comply immediately with all requirements – including watermarking and technical labeling.
Code of Practice: Technical Guidance Is Final
Alongside the Omnibus package, the EU Commission has completed the Code of Practice on Transparency of AI-Generated Content: the final version was published on June 10, 2026, with a signing deadline of July 22, 2026. The voluntary code contains, for the first time, the official EU AI icons – a standardized "AI" pictogram for uniform labeling.
The Code of Practice recommends a multi-layered approach:
- Visible labels: Unified "AI" icon for users
- Machine-readable metadata: Schema.org / IPTC / C2PA standards
- Invisible watermarks: Technical signals in images, audio, and video
- Mark preservation: Watermarks must not be removable
Why You Should Still Act NOW
The deadline extension sounds tempting – but there are three good reasons not to wait:
1. August 2026 Still Applies The general transparency obligations have not been postponed. If you publish AI content today, you must be transparent from August 2026.
2. Fines Are Substantial Violations of Art. 50 can result in fines of up to €15 million or 3% of global annual turnover.
3. Compliance Takes Time Audit logs, internal processes, and employee training cannot be implemented overnight. Start now, and you will have solid compliance documentation by August 2026.
How AIActify Prepares You
AIActify already covers all requirements becoming mandatory on both August 2 and December 2, 2026:
- ✅ Visible labels – compliant with the upcoming EU AI icon
- ✅ Schema.org metadata – machine-readable for crawlers and authorities
- ✅ Immutable audit logs – legally valid proof from day one
- ✅ Script auditor – detects prohibited AI practices (Art. 5)
- ✅ Visual editor – label AI content with a click, no code required
One script tag, 2-minute setup – and you are on the safe side.
---
Sources: EU Council – AI Omnibus Political Agreement, Latham & Watkins, Taylor Wessing, JD Supra, White & Case
