Guidelines for the Responsible Use of AI in the Public Service (Ireland)

What it is

The Guidelines for the Responsible Use of Artificial Intelligence in the Public Service are Irish Government guidance for public service bodies that are considering, buying, building or running AI. They were launched on 8 May 2025 by the then Minister for Public Expenditure and Reform, Jack Chambers, and are published by the department now called the Department of Public Expenditure, Infrastructure, Public Service Reform and Digitalisation.

The guidelines aim to support responsible innovation in a way that is practical and easy to follow, and put the human firmly in the process to build public trust in how government uses AI. They sit within Ireland’s national AI strategy and alongside the binding EU rules that already apply to public bodies.

The guidelines come with a set of supporting tools. These are a statement of the Government’s principles for responsible AI, a decision framework for evaluating whether AI should be used for a given need, a Responsible AI Canvas tool for the planning stage, and AI lifecycle guidance. They also include examples of AI already in use across Irish public services, and the Institute of Public Administration provides training on applying them.

Who it applies to

The guidelines are written for Irish public service bodies, from government departments to agencies and health service organisations. They are guidance rather than law, so they do not carry penalties. They do set the expectations against which a public body’s AI use will be judged internally and by the public.

They matter beyond the public sector in two ways. Suppliers selling AI products or services to Irish public bodies should expect procurement and due diligence questions drawn from them. And because the principles follow the EU’s seven requirements for trustworthy AI, they are a usable reference for any organisation in Europe looking for a plain, government-backed statement of good practice.

What it requires

Seven principles. The guidelines adopt seven principles for trustworthy AI, aligned with the EU’s Ethics Guidelines for Trustworthy AI. These are human agency and oversight, technical robustness and safety, privacy and data governance, transparency, diversity, non-discrimination and fairness, societal and environmental well-being, and accountability.

In practice they translate into concrete expectations. Someone in the organisation must remain accountable for each part of the AI lifecycle. Systems need strong security, because an attacked system could make different decisions or stop working. Public bodies engage their Data Protection Officer and complete a data protection impact assessment, with a contingency plan ready if the system does not perform as expected. People must be told when they are interacting with an AI system, such as a chatbot. Datasets must be checked for unfair bias, with issues resolved or mitigated. Environmental impact and effects on people’s physical and mental well-being must be considered. And the public must be able to rely on a named human being accountable for each AI solution.

Risk awareness. The guidelines describe AI uses in terms that follow the EU AI Act’s tiers, from unacceptable-risk prohibited systems and high-risk uses to limited-risk uses with transparency duties and minimal-risk uses. A public body is expected to understand where a proposed use falls.

Decide before you build. The decision framework treats AI adoption as a business case. The guidelines say the starting option should be to do nothing, and that the choice to build, buy or use a free-of-charge solution, and when, needs careful thought. Evaluation should compare the AI system with the status quo and support accountability for public spending.

Plan and run across the lifecycle. The Responsible AI Canvas structures the planning stage, and the AI Lifecycle Guidance tool supports the work that follows across the life of the system.

Key dates

The guidelines were launched on 8 May 2025. Separately, Ireland’s Regulation of Artificial Intelligence Act 2026 was signed into law on 21 July 2026. It implements the EU AI Act in Ireland, establishing Oifig IS na hÉireann, the AI Office of Ireland, as the central coordinating authority, with existing sectoral regulators supervising AI in their own areas. The Act creates no new obligations beyond the EU AI Act, and its context is covered on the EU AI Act page.

How it relates to other frameworks

For Irish public bodies, the EU AI Act and the GDPR are the binding layer. A public body deploying a high-risk AI system must, for example, carry out a fundamental rights impact assessment under the AI Act, and any solely automated decision about a person engages GDPR Article 22. The guidelines sit on top of that law and help a body apply it consistently.

The principles are the EU’s seven requirements for trustworthy AI, so the ALTAI checklist is a ready-made way to work through them for a specific system. Organisations that also follow ISO/IEC 42001 or the OECD AI Principles will find the expectations map closely, since all of them share the same core themes of accountability, transparency, fairness and human oversight.