Key Developments

The EU AI Act has officially transitioned from preparation to active enforcement mode this March, marking a critical milestone for AI regulation across Europe. The European Commission recently concluded its recruitment drive for AI technology specialists to join the European AI Office, with applications closing on March 27th. This recruitment strengthens the Office’s capacity to oversee the most advanced AI systems under the Act.

March 2026 represents the first month where General Purpose AI (GPAI) model providers face active enforcement of transparency and technical documentation obligations, rather than transitional expectations. Meanwhile, the European Parliament has formally adopted amendments aimed at simplifying the Act, signalling both implementation challenges and increasing pressure to ease compliance burdens for businesses.

Industry Context

The shift to active enforcement comes amid mixed progress on implementation support. Notably, the European Commission missed its February deadline to publish guidance on high-risk AI system obligations, potentially complicating compliance preparations for many organizations. However, the Second Draft Code of Practice on AI transparency marking and labelling was published on March 3rd, with final rules expected by June and becoming applicable in August 2026.

The introduction of the Digital Omnibus proposal in November 2025 adds another layer of complexity, potentially linking compliance deadlines to the availability of standards and support tools.

Practical Implications

For Irish and European AI developers, this enforcement phase means documentation packages must now be ready for regulatory review rather than development. GPAI model providers must maintain comprehensive technical documentation and make it available to the European AI Office upon request.

Organizations should prepare for the major August 2026 deadline when most remaining Act provisions, including demanding high-risk AI system obligations, become active. Member States, including Ireland, must also establish AI regulatory sandboxes by August 2nd.

Open Questions

The delayed guidance on high-risk AI obligations leaves uncertainty around compliance requirements. The ongoing amendment process, particularly through the Digital Omnibus, may shift timelines and obligations. How effectively the newly recruited AI specialists can manage the complex oversight landscape remains to be seen as enforcement ramps up across the European market.

Source: European Commission