29 Days to the EU AI Act: The Omnibus Changes Every Company Should Check

Twenty-nine days. That’s how long until the EU AI Act becomes fully applicable on August 2, 2026. If you sell AI-powered anything into Europe, the grace period is basically over.

Context first. The AI Act has been phasing in for a year and a half — prohibited practices and AI literacy obligations kicked in back in February 2025, and governance rules for general-purpose AI models followed in August 2025. But August 2, 2026 is the big one: the date the regulation’s full framework applies. Then, on May 7, 2026, Brussels threw a curveball. The Council and Parliament reached political agreement on the “AI Act Omnibus” — a package that clarifies requirements, extends some compliance deadlines for high-risk systems, and adds new rules on AI-generated intimate content.

Did Europe Just Blink?

Here’s the thing: the Omnibus is being read two completely different ways. Critics of the original Act call it an admission that the rules were unworkable — why else would you extend high-risk deadlines two months before full application? Supporters call it pragmatism: keep the destination, fix the road.

The most concrete change is the SME expansion. The Act’s simplified compliance framework — lighter documentation, reduced fines, regulatory sandbox access — originally targeted small and medium enterprises. The Omnibus extends it to companies with up to 750 employees and €150 million in annual revenue. That’s not a small tweak. A whole tier of mid-market European software companies just moved from the heavy lane to the light lane. Simplified guidance, standardized templates, sandboxes. Real relief.

Not everyone agrees this is wise. And honestly, they have a point. Civil society groups argue that a 750-employee company deploying AI in hiring or credit scoring poses the same risk to citizens as a 5,000-employee one. The risk didn’t shrink; the paperwork did. That tension — competitiveness versus protection — is the entire European AI debate compressed into one threshold number.

What This Means For You

If you’re a non-EU company selling into Europe — and this includes plenty of Pakistani and American SaaS firms — the Act applies to you the moment your system’s output is used in the EU. Think of it like GDPR all over again: the Brussels effect doesn’t care where your servers sit. Step one is classification. Is your system prohibited, high-risk, limited-risk, or minimal-risk? Most B2B tools land in the lower tiers, where obligations are mostly transparency. But anything touching employment, credit, education, or essential services likely sits in high-risk territory, with conformity assessments to match.

If you qualify for the expanded SME framework, use it. The sandbox access alone is worth the application — you get to test against the rules with regulator feedback instead of guessing and praying. In my experience, companies that engage regulators early spend less on compliance overall than companies that lawyer up late.

So what does this mean for the ecosystem? Watch the Dublin AI Summit on October 14, 2026. The EU is pitching it as the moment Europe reframes AI from a compliance story to a competitiveness story — heads of government, CEOs, and investors in one room. The messaging shift is already visible.

How To Get Ready in 4 Steps

One: inventory every AI system you deploy or sell, and classify each against the Act’s risk tiers. Two: check whether the new 750-employee, €150 million threshold covers you — if yes, register interest in a regulatory sandbox. Three: assign one owner for AI documentation; scattered ownership is where compliance dies. Four: calendar August 2 and re-audit against the final Omnibus text, because clarified requirements still mean changed requirements.

Key Takeaways

  • The EU AI Act becomes fully applicable August 2, 2026 — under a month away.
  • The May 7 “AI Act Omnibus” agreement extends high-risk deadlines and clarifies requirements.
  • The simplified SME framework now covers companies up to 750 employees and €150 million revenue.
  • Non-EU companies serving EU users are in scope — classification is the urgent first step.
  • The Dublin AI Summit on October 14, 2026 signals Europe’s pivot from regulation-first to competitiveness messaging.

Europe bet that trust would become a feature, not a tax. In 29 days we start finding out. Is your AI stack classified, documented, and ready — or are you hoping the deadline moves again?