Parliament Votes to Reshape AI Act Implementation

The European Parliament has approved significant changes to the EU AI Act under the Digital Omnibus package, fundamentally altering the regulatory timeline and scope. The March 26, 2026 vote opens trilogue negotiations with the Council and Commission, potentially reshaping how high-risk AI systems are regulated across the EU.

Key Timeline Extensions Provide Relief

The most substantial change postpones high-risk AI system obligations by over a year. Stand-alone high-risk AI systems now face compliance deadlines of December 2, 2027, while those embedded in products get until August 2, 2028. This represents a significant shift from the original aggressive timeline that had industry scrambling for compliance.

Equally controversial is the proposal to exclude high-risk AI systems already covered by sectoral legislation from the AI Act’s scope entirely. This could create a patchwork of regulations where AI-specific risks are addressed differently across industries.

Implementation Reality Hits Hard

The delays appear necessary given implementation challenges. With fewer than five months until most AI Act provisions take effect in August 2026, only 8 of 27 EU member states have designated enforcement contact points - a requirement that was supposed to be completed by August 2025.

For Irish companies and EU-based AI developers, this creates both opportunities and uncertainties. The extended timelines provide more preparation time, but the scope changes could leave some systems in regulatory limbo.

Practical Implications for Builders

The European Commission’s updated Code of Practice on AI-generated content marking offers more flexibility, suggesting regulators are responding to industry feedback about compliance burden. However, the potential for fines up to €35 million or 7% of global turnover remains unchanged.

Critical Questions Remain

The biggest uncertainty centres on technical safeguard assessments, which will likely become the primary enforcement battleground. How national authorities will evaluate AI system safety measures remains unclear, particularly given the limited enforcement infrastructure currently in place.

As trilogue negotiations begin, Irish and European AI companies should monitor these discussions closely while continuing compliance preparations for the August 2026 general provisions deadline.

Source: European Parliament