Libertax
AI & BORDERLESS WORK · AI TURNING POINTSINS-20260727-02

The AI Act Meets the AI Omnibus

Europe's AI rulebook reached its general application phase only after the EU had already changed important implementation timelines. That does not prove regulatory failure; it shows the difficulty of governing a technology that moves faster than its standards.

Carlos Barredo Lago4 min
Libertax editorial visualization for “The AI Act Meets the AI Omnibus”
A Libertax editorial composition about The AI Act Meets the AI Omnibus.

KEY TAKEAWAYS

KEY POINT 01The AI Act is now operational across a much broader part of its framework, but not every obligation shares the same application date.
KEY POINT 02The AI Omnibus is a material legal change, not simply political commentary. It simplifies parts of the framework and extends important timelines.
KEY POINT 03Businesses need a living compliance map rather than a single “AI Act deadline”.

The EU AI Act reached its general date of application on 2 August 2026.

Six days earlier, on 27 July, the AI Omnibus had entered into force and changed parts of the implementation framework.

That sequence captures the central problem of regulating artificial intelligence: the law needs stability, while the technology, technical standards and commercial practices continue to move.

The correct conclusion is not that the AI Act has failed.

It is that implementation risk is now part of AI regulatory risk.

Key takeaways

  • The AI Act is now operational across a much broader part of its framework, but not every obligation shares the same application date.
  • The AI Omnibus is a material legal change, not simply political commentary. It simplifies parts of the framework and extends important timelines.
  • Businesses need a living compliance map rather than a single “AI Act deadline”.

The facts as of 13 August 2026

The AI Act entered into force on 1 August 2024 and was built around staged application.

The AI Omnibus entered into force on 27 July 2026. The European Commission describes it as targeted simplification intended to improve clarity, support innovation and extend implementation timelines.

The AI Act’s general date of application followed on 2 August 2026.

Important high-risk obligations no longer follow the timetable businesses originally expected. Commission guidance identifies later dates including 2 December 2027 for certain high-risk AI systems and 2 August 2028 for high-risk AI embedded in regulated products.

Article 50 transparency obligations apply from 2 August 2026 subject to their scope and specific transitional rules.

There is no single date that answers every compliance question.

Interpretation: implementation caught up with legislative ambition

The original Act tried to create a horizontal framework before the supporting standards and market practices were fully mature.

Legal obligations can be written before businesses have detailed technical methods for demonstrating compliance. Standards then become critical because they can convert broad legal requirements into repeatable engineering and governance processes.

The Commission acknowledges delays in high-risk standardisation work.

The Omnibus is one response to that gap.

The mechanism

uncertain technical standard → uncertain compliance design → harder product and procurement decisions → delayed or duplicated investment → pressure for clarification or simplification.

The reverse is also possible. Clearer deadlines and standards can reduce uncertainty and allow firms to invest with greater confidence.

That is why postponement cannot automatically be classified either as pro-business success or regulatory failure.

The strongest countercase

Changing an implementation timetable before imposing obligations that industry cannot reliably operationalise may be evidence of adaptive regulation.

A regulator that refuses to adjust when standards are late would not necessarily be more credible.

The test is what follows: does simplification produce clearer, enforceable rules, or merely another cycle of uncertainty?

Scenarios, not forecasts

Under stable implementation, the new dates hold, standards arrive and the compliance framework becomes progressively routine.

Under further simplification, additional areas are adjusted as practical implementation exposes more friction.

Under persistent uncertainty, businesses repeatedly redesign compliance around moving dates and divergent guidance.

Practical consequences

A company should not ask only: “Does the AI Act apply to us?”

It should map each significant system separately.

What is the use case? What role does the company occupy? Does the system interact with employees, customers or the public? Is it integrated into another regulated product? Which obligations apply now, and which arise later?

The result should be a dated compliance map with owners and evidence.

For internationally active companies, regulatory geography is part of business design.

Sources

Disclaimer

This Insight reflects general information available as of the date of drafting. It is not legal or regulatory advice. The AI Act has multiple application dates, roles, exceptions and transitional provisions. Current legislation, standards and official guidance should be checked before any compliance decision.

CONTINUE READINGInsights ↗