Walk into nearly any major US tech company’s European headquarters and there is a decent chance you are standing in Dublin. Google, Meta, Microsoft, TikTok, and X all run their EU operations out of Ireland — which is exactly why the regulator Ireland builds to police AI matters far beyond its own 5.4 million citizens.
The Irish government this week approved publication of the Regulation of Artificial Intelligence Bill 2026, the legislation that will create Oifig Intleachta Shaorga na hÉireann — the AI Office of Ireland — as an independent statutory body with the power to issue compliance notices, demand documentation, and impose fines against companies that violate the EU’s AI Act. The bill does not expand what the AI Act already requires; it builds the domestic enforcement machinery Ireland needs to actually act on it.
The timing is pointed. Ireland takes over the rotating presidency of the Council of the European Union on July 1, putting the same government now standing up its own AI enforcement body in charge of steering EU-wide AI policy for the second half of 2026.
Key Developments
- Ireland’s government approved publication of the Regulation of Artificial Intelligence Bill 2026 this week, the domestic legislation implementing the EU AI Act.
- The bill creates the AI Office of Ireland, an independent statutory body that must be operational by August 1, 2026.
- Fifteen existing sectoral regulators, including the Central Bank of Ireland and the Data Protection Commission, become Market Surveillance Authorities with powers to inspect, issue compliance notices, and fine violators up to €35 million or 7% of global turnover.
- Ireland takes over the EU Council presidency on July 1, 2026, putting it in charge of steering bloc-wide AI policy at the same time it stands up its own enforcement body.
What Happened
The approval marks the latest step in a process that began more than a year ago. Ireland’s government decided in March 2025 to adopt what officials call a “distributed model” for AI Act enforcement, leveraging existing sectoral regulators rather than building a single new AI super-regulator from scratch. The Department of Enterprise, Tourism and Employment published the General Scheme — the formal heads of the bill — on February 4, 2026, and the Oireachtas Joint Committee on Enterprise, Tourism and Employment began pre-legislative scrutiny that May. This week’s Cabinet approval moves the text from a scrutinized scheme toward a bill ready for formal drafting and introduction to the Oireachtas, though it has not yet been enacted into law.
Once enacted, the bill equips Ireland’s market surveillance authorities with what officials describe as an extensive toolkit: the power to issue compliance notices, demand access to technical documentation and source code, conduct inspections, and prosecute non-compliant entities. The Competition and Consumer Protection Commission, one of fifteen designated authorities, is separately introducing a new general administrative sanctions procedure specifically to support this enforcement role.
The Mechanism: Fifteen Regulators, One Coordinator
Ireland’s distributed model is a deliberate departure from the single-regulator approach some other EU states have floated. Instead of creating one body with expertise across every domain AI touches, the bill assigns sector-specific enforcement to the regulators that already understand those sectors: the Central Bank of Ireland for AI used in financial services, the Data Protection Commission where AI systems process personal data, Coimisiún na Meán for AI in media and online platforms, and the Health and Safety Authority, the Health Products Regulatory Authority, and the Workplace Relations Commission for AI used in their respective domains, among fifteen designated authorities in total.
The AI Office of Ireland sits above that structure as a coordinator rather than a supervisor. Its statutory functions include acting as Ireland’s single point of contact with the European Commission under Article 70 of the AI Act, facilitating cooperation and shared technical expertise among the fifteen sectoral regulators, maintaining a national register of prohibited practices and high-risk systems, and operating a regulatory sandbox where companies, particularly smaller firms, can test AI systems under supervision before full market deployment. The Office does not replace any sectoral regulator’s existing powers; it exists to keep fifteen separate enforcement tracks from producing quietly inconsistent outcomes.
The Backstory
The EU AI Act itself has had direct legal effect across the bloc since entering into force in August 2024, but it required each member state to build its own enforcement infrastructure rather than relying on Brussels alone. Ireland’s government approved its first batch of eight competent authorities in March 2025, expanded that list to fifteen in September 2025 alongside a statutory instrument formally designating them, and spent the months since refining how the AI Office itself would be structured, governed, and resourced. That entire process has been running against a hard external deadline: the EU AI Act requires member states to have their national AI offices operational and their regulatory sandboxes running by August 1, 2026 — a date that does not move regardless of where any individual country’s domestic legislation stands.
That deadline is now uncomfortably close. The bill approved this week still needs full Oireachtas passage before it becomes law, and Ireland’s own AI Office is statutorily required to exist within roughly six weeks of this announcement. Ireland is not alone in racing the clock: the Brussels-level response to similar timeline pressure across the bloc has been the Digital Omnibus package, which the European Commission proposed in November 2025 specifically to push back some of the AI Act’s harder deadlines, including delaying high-risk obligations under Annex III from August 2026 to no later than December 2027. Ireland’s domestic bill is tightening enforcement capacity at almost the exact moment Brussels is loosening some of the substantive obligations that capacity will eventually be used to enforce.
Reactions
Minister for Enterprise, Tourism and Employment Peter Burke described AI as a transformative technology that nonetheless requires appropriate oversight, and said the bill gives Ireland’s competent authorities the investigative and sanctions tools they need while fulfilling the country’s EU obligations. Minister of State for Trade Promotion, Artificial Intelligence and Digital Transformation Niamh Smyth framed the legislation in broader terms, calling the AI Office’s creation a step toward building institutional foundations for AI that works for people ethically and accountably, and noting Ireland’s position as home to many of the world’s leading foundational AI model providers.
The Irish Council for Civil Liberties welcomed the underlying General Scheme back in February, saying its initial analysis found that most of its recommendations had been incorporated, and singling out the proposal for the AI Office to be a statutory, independent authority as a particularly important safeguard. Legal commentators tracking the bill have separately described its enforcement toolkit as extensive by Irish regulatory standards, comparable to powers already used in product-safety and competition enforcement rather than anything novel to AI specifically.
The Dispute: An Untested Office Meets a Real Stress Test
The AI Office of Ireland’s actual enforcement reach is about to face a test case that has nothing to do with anything Ireland controls. In May, a US export control order forced Anthropic to suspend its Mythos and Fable AI models worldwide, cutting off access for European users along with everyone else, regardless of what the EU AI Act says about those users’ rights to the service. Legal experts have already begun asking, in light of that episode, whether the AI Act’s enforcement architecture — built primarily around obligations on providers and deployers operating inside the EU — has any real mechanism for responding when access to an AI system is cut off by a foreign government’s export controls rather than by EU-side non-compliance. Ireland’s new Office inherits that open question along with its enforcement powers.
There is a second, more procedural tension in the bill’s own timeline. The AI Office is required to be statutorily operational by August 1, 2026, a deadline set at the EU level that has not moved despite the Digital Omnibus relaxations elsewhere in the Act. Yet the bill that creates the Office, having only just received Cabinet approval for publication, still needs to clear pre-legislative scrutiny, formal drafting, and a full Oireachtas vote before it has legal force. Ireland has hit earlier AI Act milestones — its competent-authority designations and single point of contact — on schedule through secondary legislation rather than a full Act of the Oireachtas, and officials have not yet said whether they would lean on a similar mechanism if the primary legislation itself runs past August without final enactment.
What Happens Next
The bill now moves toward formal drafting and introduction to the Oireachtas, where it will need to pass before the AI Office of Ireland can be formally established as a statutory body. In parallel, the existing fifteen competent authorities continue exercising the powers already assigned to them under last September’s designation, regardless of whether the new bill clears the Oireachtas before August 1. The clearest near-term marker to watch is whether the AI Office reaches operational status on schedule, and whether Ireland’s incoming EU Council presidency, beginning July 1, uses its agenda-setting position to push other member states toward similar enforcement timelines or instead spends its political capital smoothing over the Digital Omnibus delays already in motion.
Why It Matters
Because so much of the AI industry’s European footprint runs through Dublin, the AI Office of Ireland is positioned to become a de facto first line of EU AI Act enforcement for a disproportionate share of the global AI industry, regardless of how the office’s formal remit is worded. Other member states without Ireland’s concentration of Big Tech EU headquarters are watching how Dublin builds this institution, both as a model for the distributed-authority approach and as an early signal of how aggressively, or cautiously, AI Act enforcement actually gets applied once the legal infrastructure is finally in place. Whether Ireland’s answer is regulatory leadership or a cautionary tale about racing a deadline will likely become clear well before the next full review of the EU AI Act itself.
Sources
Department of Enterprise, Tourism and Employment (gov.ie); Silicon Republic; KPMG Ireland; William Fry; Irish Legal News.