AI Regulation Expert & Founder of AIAct-Akademie
AI REGULATION | TRILOGUE AGREEMENT
AI Omnibus:
EU Postpones AI Act Deadlines for High-Risk AI
The European Commission, Parliament, and Council agreed on sweeping changes to the AI Act on the night of May 7, 2026 – the so-called 'AI Omnibus'. The deadline for high-risk AI systems is pushed back by 16 months, industrial machinery is largely exempted, and new bans on nudifier apps and deepfake pornography are set to take effect. The provisional agreement still requires formal adoption by Parliament and the Council.
Key Takeaways
- EU trilogue agreement on May 7, 2026: The 'AI Omnibus' amends the AI Act.
- High-risk AI deadline postponed: From August 2026 to December 2027.
- Industrial machinery largely exempted from high-risk requirements.
- New bans: Nudifier apps and non-consensual deepfake pornography from December 2026.
- Watermarking requirement for AI-generated content incoming.
What Has Changed?
The Omnibus Legislative Package VII – the 'AI Omnibus' – brings four key changes to the EU AI Act:
| Change | Before | New |
|---|---|---|
| High-risk AI obligations | August 2026 | December 2027 |
| Industrial machinery | High-risk rules applied | Largely exempted |
| Nudifier apps / Deepfakes | Not explicitly regulated (partly covered by Art. 5) | Banned from December 2026 |
| Watermarking AI content | Not regulated | Required from December 2026 |
The deadline extension for high-risk AI affects systems in biometric surveillance, critical infrastructure, and law enforcement. Companies gain 16 additional months of preparation time. (Full text of the AI Regulation)
At the same time, the EU tightens other areas: Nudifier apps and non-consensual deepfake pornography are explicitly banned. AI-generated content must be tamper-proof watermarked in the future.
Industry Reactions
The agreement is broadly welcomed by business, but industry representatives say it does not go far enough. According to Bitkom, the compromise avoids duplicate testing and documentation processes, allowing companies to invest resources more specifically in innovation.
The agreement on the AI Omnibus is a positive signal for the development of industrial AI in Europe.
Dehmel also calls for the approach adopted for machinery to be extended to other highly regulated sectors such as medical devices and radio equipment. Industry voices also warn that overly rigid regulations could make it harder to compete with providers from the US and Asia. (Bitkom press release)
What Does This Mean for Companies?
High-Risk AI
High-risk AI providers gain 16 additional months for compliance preparation. The new deadline is December 2027 instead of August 2026.
Machinery Manufacturers
Machinery manufacturers benefit from the broad exemption of their products from high-risk requirements.
New Bans
All AI providers must plan for the new bans on nudifier apps and deepfake pornography, as well as the watermarking requirement from December 2026.
Generative AI
For generative AI systems, transparency deadlines remain unclear – the promised interpretive guidance is still pending.
Conclusion: More Time, But No Free Pass
The AI Omnibus gives companies breathing room on high-risk compliance – but the 16 months gained are not an invitation to wait. The core obligations of the AI Act remain in place, the AI competence requirement under Art. 4 has been in effect since February 2025, and the new bans on nudifier apps and deepfakes take effect as early as December 2026. Companies should use the time gained to systematically build compliance processes. The pending interpretive guidance will show how strict the requirements actually become in practice.