AI Regulation Tracker / Legislation pending
Ireland's AI Bill would name enforcers and adopt EU 7% turnover fines
Ireland's Regulation of Artificial Intelligence Bill 2026 would split AI supervision across existing regulators and import the EU AI Act penalty ceiling of 7% of global turnover. It is still moving through the Seanad, not yet law.
Ireland has set out how it intends to police artificial intelligence, and the answer is not a single new agency. The Regulation of Artificial Intelligence Bill 2026, published on June 17 by the Department of Enterprise, Tourism and Employment and running to roughly 139 sections, gives domestic effect to the EU AI Act by spreading supervision across regulators that already exist. The Bill has moved into the Seanad and is not yet law, so its provisions describe what is coming rather than what binds today.
Who would enforce it
Rather than concentrate authority, the Bill designates a set of existing bodies as market-surveillance authorities for AI. Among the named regulators are the Data Protection Commission, which handles fundamental-rights questions tied to personal data, the Central Bank of Ireland for regulated financial services, and the Competition and Consumer Protection Commission, which is adding an administrative-sanctions procedure to its powers under the Competition and Consumer Protection Act 2014. Reporting on the Bill indicates around thirteen authorities in total, drawing in sectoral regulators for media, utilities, health, and employment. A central coordinating body, the AI Office of Ireland, would sit above them.
For a licensed professional, the practical point is that there is no one desk to call. A bank deploying a credit-scoring model answers to a different supervisor than a hospital using triage software or a media platform running recommendation systems. The AI Office of Ireland is intended to coordinate across these bodies and act as the central point for implementing the EU regime, but the day-to-day supervision of a given AI system would sit with whichever regulator already owns that sector. That design choice keeps enforcement close to specialist knowledge, and it also means a single company can face more than one Irish authority depending on how many kinds of AI it runs.
The penalties
The Bill adopts the EU AI Act penalty regime. The most serious tier, for prohibited practices under Article 5 of the AI Act, carries fines up to 7% of total worldwide annual turnover, or 35 million euros, whichever is higher. Lower tiers apply to other breaches. The 7% figure matters because it is calculated on global revenue, not Irish revenue, so a large multinational supervised through an Irish entity faces a ceiling tied to its worldwide business.
The Bill pairs these sanctions with a graduated toolkit. Enforcement is designed to start with cooperative measures such as compliance notices, then escalate where necessary through prohibition orders and seizure powers, and ultimately to administrative fines and criminal prosecution. Independent adjudication and court oversight are built into the process, which means the heaviest penalties are not imposed by a regulator acting alone. For most firms, the realistic exposure begins with notices and information requests rather than headline fines, but the top tier sets the outer limit of what non-compliance can cost.
What it does not do
The Bill does not create new substantive AI rules of its own. The obligations, prohibitions, and risk classifications come from the EU AI Act, which applies across the bloc. What the Irish legislation supplies is the enforcement architecture: who supervises, what powers they hold, and how penalties are imposed. It also does not take effect on publication. Because it is still before the Seanad, the designations and fine ceilings are proposed, not operative, and the text could change before enactment.
The cross-border angle
This matters well beyond Ireland. Dublin is the EU base for many large US technology and financial firms, which means a US company running AI systems from an Irish entity would answer to these named Irish regulators for AI Act compliance. Ireland's distributed model is also instructive on its own terms. Where some jurisdictions are building a single AI regulator, Ireland is layering AI supervision onto the regulators that already know each sector. For US compliance officers watching how the EU AI Act lands in practice, the Irish approach previews the kind of multi-regulator enforcement that firms operating in Europe will need to plan around.
The immediate task is preparation, not filing. Firms can inventory the AI systems they run in Ireland, map each to its probable supervising authority, and confirm whether any use could fall into the prohibited category that draws the heaviest fine.
Frequently Asked Questions
What does the Regulation of Artificial Intelligence Bill 2026 change?
It gives Ireland the machinery to enforce the EU AI Act, naming existing regulators as market-surveillance authorities and adopting fines up to 7% of global annual turnover for prohibited practices. It does not add new AI rules beyond the EU Act.
Who does it affect?
Providers and deployers of AI systems operating in Ireland, including US firms headquartered in Dublin, across finance, data protection, consumer, media, utilities, health, and employment.
Is it in force yet?
No. It was published on June 17, 2026 and is progressing through the Seanad. It is a bill in progress, and no commencement date has been set.
Which regulator would supervise our AI system?
It depends on sector. The Bill uses a distributed model, so a financial-services model likely falls to the Central Bank, personal-data questions to the Data Protection Commission, and consumer matters to the Competition and Consumer Protection Commission, coordinated by the AI Office of Ireland.
Sponsored Training
Browse the full AI Regulation News tracker
Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.