Fewer than half of Government departments and agencies had facilitated training in artificial intelligence systems at the onset of 2025 as the organs of the State attempt to keep pace with the rapidly advancing technology and its potential applications.
Detailed responses to questions from 18 Government departments and 40 other official bodies at the end of January show a widely varied patchwork of approaches, with some embracing the technology and others avoiding its use, at least for now.
Just slightly more than half indicated the existence of, or plans for, tailored policy.
Last week, the Department of Public Expenditure published updated guidelines on the ethical use of AI by civil servants. It has launched resources designed to support the adoption of AI, including a framework for evaluating potential use, and freely available courses for staff.
Mexican beauty influencer Valeria Márquez shot dead while livestreaming on TikTok
PGA Championship Round 1 live updates: Rory McIlroy and Shane Lowry among early starters
Galway’s persistently high rents to be investigated by RTB following ‘concerning’ trends in new data
Look inside: Former RTÉ presenter’s Glenageary home with French countryside-style garden for €2.25m
“AI offers immense possibilities to improve the provision of public services,” Minister for Public Expenditure Jack Chambers said. “It is crucial that we embrace that change and adapt quickly in order to deliver better policy.”
To date, however, many agencies have relied solely on broad guidance documents circulated in recent years and have not yet tailored individual policies, explored available technology or implemented training programmes.
A breakdown of responses issued to Malcolm Byrne, the Fianna Fáil TD set to chair Ireland’s nascent Oireachtas committee on AI, shows training has been provided or planned for in half of Government departments, and in just slightly more than half of State agencies.
Individual, targeted policies, meanwhile, exist or are planned for in about a third of departments and in about 60 per cent of sub-agencies responsible for specific aspects of the State bureaucracy. Fourteen Government departments specifically ruled out the use of AI in the design or formulation of legislation, although there is no indication of its use to do so in any.
‘I’ve entrepreneurial spirit in my veins’ – Apprentice star Jordan Dargan
Although not all departments indicated individual policies, many reference their adherence to broader documents issued in recent years. These include guidance from the National Cyber Security Centre (NCSC), taking in generative artificial intelligence (GenAI) which produces original output in response to prompts from users.
The National AI strategy and interim guidelines for the public service, issued in January 2024, are also referenced, while the EU Artificial Intelligence (AI) Act, which came into force in August 2024, has implications for policy formation.
Mr Byrne said he hoped the new guidelines for the Civil Service would prompt a shift in thinking.
“There were certainly some positive examples [of AI use in January’s responses] . . . but I was a bit concerned that it was very much an abundance of caution, ranging from ‘do not use AI in any circumstances’ to ‘well, we might see some use in the future’,” he said.
“My view is that public services should be using AI for the purposes of efficiency gains . . . to remove mundane tasks so that large swathes of the public service can spend more time focusing on policy issues.”
[ Artificial intelligence to be used in drafting Government department documentsOpens in new window ]
As well as the provision or lack of training and policy, responses from 58 organisations offer a snapshot in time as to how AI is viewed, variously, by officials.
The Department of Finance, citing NCSC advice, told Mr Byrne that GenAI ought not to be used to generate responses to correspondence such as parliamentary questions or public representations, or in designing or drafting policy.
Staff at the Garda National Data Protection Office have undergone training in assessing the risks of AI systems in relation to data protection issues. The Agri-Food Regulator said the use of GenAI was not permitted, a policy in line with the Department of Agriculture. Similarly, the National Transport Authority (NTA) said the use of GenAI was prohibited “by default”, while recognising its potential benefits.
Elsewhere, there are signs of the technology being embraced. The Loughs Agency, which protects fisheries, said it was developing AI for use in validating its fish counters. The HSE has set up an implementation working group for the use of AI in the public health sector.
Unlike the majority of agencies, Teagasc, the agriculture and food development authority, said it uses large language AI models, while acknowledging this “presents risks that must be managed”.
