Last week, the European Parliament approved the AI Omnibus, a major update to the EU AI Act. This package of amendments delays some deadlines, bans nudifier apps, and tweaks how high-risk systems are handled. The goal is simple: make AI compliance more practical while keeping strong protections in place.
What Is the AI Omnibus?
The name sounds complex, but the concept is straightforward. The Omnibus is a set of targeted changes to the AI Act designed to make implementation more realistic. It has three main goals: align deadlines with the availability of standards and support tools, extend SME simplifications to slightly larger companies, and streamline governance by giving the EU AI Office more oversight over complex, cross-border systems.
New Deadlines for High-Risk AI
MEPs backed fixed new application dates for high-risk obligations. Stand-alone high-risk systems like AI used in employment, education, essential services, or migration will have to comply from 2 December 2027. AI used as a safety component in products already covered by EU safety law (think machinery, medical devices, toys) will have to comply from 2 August 2028.
Parliament also agreed to delay watermarking rules for AI-generated content. Those obligations now start on 2 December 2026. From that date, AI-generated content must be labelled in a machine-readable way so that other systems can detect it.
Ban on Nudifier Apps
Parliament and Council agreed to ban AI systems that create child sexual abuse material or depict intimate parts or sexually explicit activities of an identifiable person without consent. These are known as ‘nudifier’ apps, and they’re now explicitly prohibited.
Narrowing the High-Risk Classification
One earlier concern about the AI Act was over-classification, the risk that almost anything containing AI might be treated as high-risk. The Omnibus narrows the definition of a “safety component.” It clarifies that AI features which merely assist users or optimise performance, and whose failure does not create health or safety risks, are not automatically treated as high-risk.
Bias Detection and Data Use
Parliament approved language that allows personal data to be processed, where strictly necessary and with safeguards, to detect and correct bias. This can apply to both high-risk and some non-high-risk systems. The EU is trying to balance two things: strong data protection and the reality that you often need demographic and outcomes data to see whether your model is discriminating and to fix it.
Support for SMEs and Innovators
The original AI Act already included lighter obligations and support measures for SMEs. The Omnibus extends some of those simplifications to “small mid-cap” companies, firms that are bigger than SMEs but not yet large corporates. We also see more emphasis on regulatory sandboxes. The Omnibus broadens access so more innovators can test AI systems in real-world conditions under supervision, and it sets up an EU-level sandbox from 2028, with a focus on core industries like automotive.
Stronger EU AI Office Powers
The Omnibus strengthens the powers of the AI Office, the new central enforcement body in Brussels. The Office gets a stronger role over some general-purpose AI systems and AI deeply embedded in large platforms and search engines. The idea is to reduce fragmentation, so companies aren’t dealing with 27 different interpretations of the rules for systems that operate across borders by design.
Part of a Wider Simplification Push
Beyond AI, the Omnibus is part of a broader attempt to tidy up overlapping digital rules. On cybersecurity, it moves towards a single entry point for incident reporting. Right now, a serious incident might need to be reported under NIS2, GDPR, DORA and other frameworks, often in different formats to different authorities. The goal is one interface where companies can meet those obligations more efficiently.
On privacy, the Commission is proposing targeted tweaks to GDPR to harmonise and clarify certain rules for AI and data-driven innovation, while keeping core protections intact. Politically, the message is clear: Europe is keeping its high standards, but is willing to adjust the plumbing so that AI innovation and compliance are actually workable.
Â
If you want to build your AI literacy and skills, I can help
📚 Grab AI Playbook on Amazon, the perfect podcast companion.
📰 Subscribe to Human in the Loop, my LinkedIn Newsletter
🎓 Take the CPD Accredited AI Playbook Diploma
📞 Book a call at digitaltraining.ie
Sources
European Parliament: Press release on the AI Omnibus approval. https://ec.europa.eu/commission/presscorner/api/files/document/print/en/ip_26_1024/IP_26_1024_EN.pdf
European Commission: Details on the AI Omnibus and simplification measures. https://ec.europa.eu/commission/presscorner/detail/en/ip_25_2718
Key Statistics
– Stand-alone high-risk AI systems must comply from 2 December 2027.
– AI used as a safety component in products must comply from 2 August 2028.
– Watermarking rules for AI-generated content start on 2 December 2026.
– An EU-level regulatory sandbox will be set up from 2028, with a focus on core industries like automotive.
– Companies currently may need to report serious incidents under multiple frameworks including NIS2, GDPR, and DORA.
– The AI Office will oversee systems operating across 27 member states to reduce fragmentation.
About the Podcast Host
AI and Communications Specialist | Author | Podcaster | International Speaker
Joanne Sweeney is a recognised leader in digital marketing and digital transformation, specialising in the practical application of AI to enhance communication strategies. The founder of Digital Training Institute, Joanne, empowers leaders and communications professionals to leverage AI responsibly and effectively, ensuring transparency, trust, and tangible results.
She completed the Harvard Kennedy School Executive Programme in AI and Public Policy this summer before launching her fourth book AI Playbook, in September. Joanne has also published two editions of Public Sector Marketing Pro, which were preceded by her first title, Social Media Under Investigation Law Enforcement and the Social Web.
Her books have become go-to resources for governments, public sector and policing agencies, as well as marketing and communications professionals, all navigating the challenges of the changing Internet.
Joanne is the host of the AI SIX Podcast, a daily six-minute podcast that provides actionable insights, expert interviews, and practical advice to help professionals navigate the new world of work.
