The Trilogue Breakdown: What Just Happened

On 28 April 2026, the second political trilogue between the European Parliament, Council of the EU, and European Commission collapsed without agreement. The sticking point: whether AI systems embedded in products already regulated by existing EU sectoral rules—medical devices, industrial machinery, toys, connected cars—should face additional AI Act requirements or be governed solely by sectoral frameworks.

This wasn’t a minor technical dispute. The intersection of the AI Act with pre-existing safety regimes represents one of the most consequential implementation questions facing European enterprises. And with Ireland taking over the EU presidency on 30 June 2026, just two months away, the timing couldn’t be more fraught.

Why Ireland Is Under Particular Pressure

Ireland faces a unique enforcement burden. As home to EU headquarters for most major tech companies, Ireland handled 60% of the €250 million in GPAI-related fines issued across EU member states in Q1 2026 alone. This pattern suggests Ireland’s Data Protection Commission and emerging AI Office will inherit the enforcement lion’s share when high-risk AI rules activate in August.

Yet Ireland must operationalise its AI Office by 1 August 2026—just 95 days away—while simultaneously establishing at least one national AI regulatory sandbox. These deadlines are non-negotiable under the AI Act. But if trilogue negotiations drag on until August, Ireland’s presidency could be consumed by salvage diplomacy rather than strategic implementation.

The Cascade of Practical Chaos

For enterprises building or deploying high-risk AI systems, the current limbo creates three converging problems:

1. Sectoral Exemption Uncertainty: Medical device manufacturers, automotive suppliers, and industrial automation providers don’t yet know whether their AI components face dual compliance regimes or streamlined sectoral pathways. This affects design, documentation, and validation timelines.

2. Sandbox Readiness Gaps: Member states are racing to establish sandboxes by 2 August 2026, but without clarity on exemption scope, these sandboxes may lack coherent standards. Irish innovators competing in EU markets need regulatory certainty, not moving targets.

3. Ireland’s Enforcement Identity Crisis: If the trilogue remains unresolved, Ireland’s AI Office launches with ambiguous mandate boundaries. Does it enforce the full AI Act, or negotiate case-by-case exemptions? How do 15 Irish enforcement authorities coordinate?

The Path Forward (and What Happens If It Fails)

Sources indicate a “potentially clear path” toward simplified agreement ahead of mid-May trilogue discussions. If successful, enforcement rules could crystallise before August.

But if no deal emerges by August, the original deadline stands. This means high-risk AI system enforcement activates regardless of sectoral exemption clarity—forcing enterprises into conservative compliance postures and potentially fragmenting European markets across different national interpretations.

Ireland’s presidency offers a diplomatic window. But it also means Irish regulators will be negotiating European consensus while simultaneously standing up their own enforcement infrastructure.

Open Questions

  • Will May trilogue talks produce binding positions, or merely extend uncertainty?
  • How will Ireland’s AI Office coordinate with its Data Protection Commission on overlapping jurisdictions?
  • Which sectoral rules will definitively “trump” the AI Act, and which will run parallel?
  • Will sandboxes launched before clarity actually serve their purpose, or become symbolic exercises?

For Irish and European AI builders: prepare dual compliance timelines. The August deadline is not moving. But what you’re complying with may still be in flux.

Source: artificialintelligenceact.eu