---

State use of AI to retain human oversight

Guidelines for responsible AI use in the public sector have been launched by the Government.

They outline the acceptable and unacceptable uses of AI for Government departments, state bodies, and ongoing public-sector AI projects.

The largest public-sector trade union, Fórsa, said that there were challenges ahead, with risks to various types of jobs – including skilled jobs.

Public-sector employees can use AI to generate draft documents, summarise reports, and translate text.

AI can also create new content based on input data.

In healthcare, AI can identify and categorise data, analyse documents, detect anomalies, and support disease diagnosis.

Public Expenditure Minister Jack Chambers clarified that broadening AI use did not reduce public-service numbers.

The plan emphasises safeguards, human oversight, and dependable, secure AI systems.

Systems must have robust policies to protect personal data and highlight transparency and upfront communication with end-users.

The plan also requires fair and inclusive AI systems. Revenue currently uses large language models (LLMs) to route taxpayer queries more efficiently.

Broad use

Minister Chambers stated that, while AI was fragmented in the public sector, these guidelines aimed to encourage and empower public servants to use it more broadly.

The guidelines recommend human oversight of AI systems in the public service.

The guidelines advise against the use of popular free generative AI (GenAI) models – including popular chatbot tools such as ChatGPT and DeepSeek.

The document states that technology and the associated regulatory frameworks for AI and Gen AI are advancing rapidly.

“To remain effective and relevant, these guidelines are designed as a living document, adaptable to ongoing changes and emerging best practices,” it says.