On 14 October 2026, Ireland's AI Office advanced plans for its AI Regulatory Sandbox Programme and a National AI Literacy Taskforce. Budget 2027 added €2.3m, bringing the office's budget to €3.8m, according to Silicon Republic. The announcement is mostly about AI. It also bears on a harder question: how Ireland's overlapping digital regulators, including Coimisiún na Meán, the media regulator created under the Online Safety and Media Regulation Act 2022, avoid giving companies conflicting answers.
What was announced
The sandbox was first announced in Ireland's refreshed 2024 AI strategy. It is meant to be a controlled environment where AI providers and regulators can clarify their roles and assess compliance with the EU AI Act, GDPR and the Digital Services Act. Participants follow an agreed Sandbox Plan that sets regulatory objectives, timelines and milestones. Engagement includes regulatory discussions, documentation review and, in some cases, specialist technical assessments. AI Office CEO Paul Byrne called establishing the sandbox a key priority and part of meeting the office's obligations under the AI Act. The AI Literacy Taskforce is to deliver a national framework by the end of 2027.
The sandbox exists because EU law requires it. Article 57 of the AI Act obliges each Member State to have at least one national AI regulatory sandbox. It says authorities "shall cooperate with other relevant authorities" where appropriate. It also provides that participants who follow the sandbox plan and act in good faith on the authority's guidance face no administrative fines for infringements of the Regulation. They remain liable for damage to third parties.
The case for caution
The strongest argument for regulatory scepticism is that a sandbox can become a comfort blanket. A regulator that blesses a plan inside a sandbox may still lack authority over other laws, and a firm can leave with false assurance. Consumer and civil-liberties groups also worry that a funded, friendly channel for providers favours well-resourced incumbents over affected users. Those concerns are legitimate, and the Article 57 fine protection only covers infringements of the AI Act itself.
Why the Coimisiún na Meán link matters
The published scope of the Irish sandbox names the AI Act, GDPR and the Digital Services Act. It does not name the Online Safety and Media Regulation Act. But the regulators are connected. The 2022 Act, signed on 10 December 2022, lets the Media Commission make binding Online Safety Codes for designated online services. It also made the Commission Ireland's Digital Services Coordinator under the DSA, according to law firm A&L Goodbody's summary. A company testing a recommender system or generative feature under the DSA is therefore dealing with the same regulator that enforces Ireland's online safety rules.
Coimisiún na Meán's Online Safety Code applies to video-sharing platform services. It covers harmful and illegal video content, audiovisual commercial communications and user-generated comments linked to videos. Its measures include age assurance, parental controls, content rating and user reporting tools. A video platform using AI to rank, moderate or generate content faces overlapping questions. Is the system a DSA risk issue, a GDPR lawful-basis issue, an AI Act classification issue, or a duty under the Online Safety Code?
A sandbox that answers only some of those questions leaves the hardest part unresolved. The risk is not that Ireland has too few regulators. It is that each gives a correct answer to a different question, and the company has to reconcile them.
What proportionate design looks like
The pro-innovation case for the sandbox is strong. Start-ups and SMEs cannot afford to guess what four regimes mean for one product. Ireland also benefits from the fact that many global platforms are established there. Legal certainty is an economic asset as well as a safeguard. The same week, IDA Ireland's Labour Market Pulse ranked Ireland fourth globally for AI engineering talent concentration. Regulatory clarity is how that talent turns into companies that stay.
Three design choices would make the sandbox worth its budget.
- Publish outcomes. Anonymised summaries of what regulators agreed in each Sandbox Plan, where doing so is lawful, would turn private assurances into guidance that non-participants can use. Without that, a €3.8m office serves a handful of firms.
- Bring the media regulator in early. Where a test involves user-generated content, video or minors, Coimisiún na Meán should be at the table as a participant, not a recipient of a report. That keeps online-safety expectations from being discovered after a product has launched.
- Protect lawful speech in the process. Cooperation among regulators must not become a route for informal pressure on content decisions. Any sandbox discussion of moderation systems should rest on evidence of measured error rates and proportionate outcomes, not on a presumption that more removal is safer.
What to watch
Silicon Republic's report does not say which regulators will take part or when the first cohort will start. Those details will determine whether this becomes a real coordination mechanism or a well-funded AI Office programme running alongside the others. The €2.3m increase is modest next to the scale of the task. The test is not the size of the budget. It is whether a founder who finishes the sandbox can tell, with confidence, what all of Ireland's regulators expect of the product.