Today in Brussels: A Major Vote on the AI Act
Thursday 26 March 2026 is a significant day for EU AI regulation. The European Parliament's full plenary is voting today on its official negotiating position on the AI Act simplification package — formally known as the Digital Omnibus VII.[1]
The package was already backed by MEPs in committee on 18 March — by a decisive 101 votes to 9, with 8 abstentions. Today's plenary vote formally adopts Parliament's mandate, clearing the way for trilogue negotiations with the Council to begin. Here's what's in it and why it matters for SMEs.
This article is for informational purposes only and does not constitute legal advice.
The Headline Change: High-Risk AI Deadlines Pushed Back
The most consequential change in the package is the postponement of compliance deadlines for high-risk AI systems. MEPs are voting to set two new fixed dates:
- 2 December 2027 — for high-risk AI systems listed directly in the AI Act (biometrics, employment decisions, education, essential services, law enforcement, border management)
- 2 August 2028 — for AI systems embedded in products already covered by EU sectoral safety laws (medical devices, machinery, radio equipment, toys)
The original deadline for these obligations was 2 August 2026. The reason for the extension: key technical standards that companies need to actually demonstrate compliance still aren't finalised. Forcing businesses to comply against an undefined standard wasn't workable — so the EU is setting fixed dates tied to when the standards are realistically expected to land.
Important caveat: these new dates are not final until trilogue concludes and the amended text is officially published. The August 2, 2026 deadline remains on the books until then. Don't use this as a reason to stop compliance work — especially on transparency obligations and AI literacy, which are unaffected.
The New Ban: AI Nudifier Apps
One of the most striking additions MEPs are voting on today is a new prohibition not in the Commission's original proposal: a ban on so-called "nudifier" AI systems.
These are AI tools that generate or manipulate images to make an identifiable real person appear nude or in sexually explicit situations — without their consent. The ban would apply to any AI system whose primary function is to enable this, unless the system has effective built-in safeguards that prevent users from creating non-consensual content.
This is a direct legislative response to widely reported harms from these tools, particularly against women and children. For most businesses, this prohibition is simply background context — if you're not in this space, it doesn't affect you. But it signals that the EU is willing to add new prohibitions to the AI Act as harmful use cases emerge.
Watermarking: A Shorter Extension Than Expected
The Commission proposed extending the deadline for AI-generated content watermarking (labelling AI-created audio, images, video, and text to indicate their origin) until 2 February 2027. MEPs are voting to shorten that extension to 2 November 2026.
If your business publishes AI-generated content at scale — marketing images, synthetic voice recordings, AI-written copy presented as human-authored — this is worth monitoring. The watermarking requirement could arrive sooner than some had anticipated.
SME Support Extended to Mid-Caps
Currently, the EU AI Act's support measures for smaller businesses — simplified documentation, lighter regulatory sandbox access, reduced conformity assessment requirements — apply only to SMEs (under 250 employees, under €50M turnover).
Today's vote backs extending these benefits to small mid-cap enterprises (SMCs): companies that have grown beyond SME thresholds but aren't yet large multinationals. This is a practical recognition that compliance burdens don't suddenly become manageable the moment you cross the 250-employee line.
A Separate Development: The EU AI Office Prepares 11+ Guidance Documents
Alongside today's vote, the European Commission announced earlier this month that the EU AI Office is preparing a comprehensive set of guidelines to support businesses in implementing the AI Act. These include:[2]
- Practical guidance on high-risk AI classification
- Templates for fundamental rights impact assessments
- Guidance on transparency requirements (Article 50)
- Guidelines specifically for SMEs on simplified quality management systems
- Joint guidelines with the European Data Protection Board on how the AI Act interacts with GDPR
All of these are expected throughout 2026. In the meantime, businesses can use the AI Act Service Desk — a free Commission resource for compliance questions.
What Happens Next
After today's plenary vote, the formal trilogue process begins — negotiations between Parliament, Council, and Commission to agree a final text. Given that the Council (which voted on 13 March) and Parliament's committees agreed on the core elements with a similar mandate,[3] there is room for optimism that negotiations will move quickly. But "quickly" in EU terms typically means months, not weeks.
What This Means for Your Business Right Now
Here's the practical guidance for SMEs:
- Transparency obligations (chatbot disclosures, AI content labelling): Still due August 2, 2026. No change. Act now.
- High-risk AI conformity assessments: Timeline may extend to December 2027 or August 2028 once the Omnibus is finalised. But the text isn't law yet — treat August 2026 as your working deadline until told otherwise.
- AI literacy (staff training): This was required since February 2025. A light-touch team briefing per AI tool is all most SMEs need to demonstrate good faith compliance.
- AI-generated content labelling: Watch the watermarking deadline — it may land as early as November 2026 if today's Parliament position holds through trilogue.
The EU is clearly moving to make compliance more achievable for smaller businesses. But the trajectory is toward more regulation, not less — just simpler, better-timed regulation. The best thing an SME can do today is understand exactly where it stands.
This article is for informational purposes only and does not constitute legal advice.
Sources
Know your EU AI Act risk level in 10 minutes
Our free audit classifies every AI system you use and tells you exactly what to do before August 2, 2026.
Start Free Audit →