EU AI Act Omnibus discussions & Documents
Over the weekend I noticed something interesting on LinkedIn.
Several posts about the EU AI Act Omnibus discussions started circulating — many sounding rather alarmist.
Some suggested organisations should expect major regulatory uncertainty or even delay AI governance efforts.
My take: This creates more confusion than clarity.
Here’s why the discussion in Brussels does not fundamentally change what organisations should already be doing today.
🎯 The AI Act in reality
The strictest obligations apply primarily to high-risk AI systems (roughly 10–15% of AI applications).
Typical examples:
- AI for recruitment/employee evaluation
- Credit scoring & financial decisions
- AI in medical devices
- Critical infrastructure decision support
- Certain education/public administration tools
By contrast: Most tools in use today — copilots, productivity assistants, analytics, generative AI features — fall into much lower risk categories.
⚠️ The real Omnibus challenge
Most will NOT face heavy regulatory obligations overnight. The Omnibus actually delays high-risk rules (to Dec 2027/Aug 2028) and extends SME simplifications.
But they need to understand where AI already exists inside their organisation.
When companies look closely, they often discover:
- AI tools introduced independently by departments
- Generative AI embedded in software platforms
- External AI services in workflows
- Decision-support systems without governance
The AI Act doesn’t create AI governance.
It reveals that many organisations haven’t structured it yet.
🚀 Practical starting point — start TODAY
Waiting for „perfect regulatory clarity“ won’t solve this. AI literacy shifts from obligation to promotion by Commission/MS.
Do this now:
✅ Create transparency over AI systems in use
✅ Establish basic governance responsibilities
✅ Define oversight & documentation structures
✅ Identify potential high-risk applications
Organisations building this early will transition smoothly.
Those waiting may find the real challenge was never the law — it was the organisation itself.
Curious to hear from you:
Are organisations already mapping their AI systems — or still waiting for regulatory clarity?
What’s the biggest „hidden AI“ discovery you’ve made so far? 👇