Ireland AI Office warns employers on staff training

Ireland's new AI Office has put employers under immediate scrutiny over how staff use artificial intelligence at work. The requirement applies to organisations using AI under the EU AI Act.

At the centre of the change is Article 4 of the bloc's AI law, which requires organisations to ensure a sufficient level of AI literacy among staff and others using AI systems on their behalf. In practice, employers must be able to show that workers have appropriate training, oversight and an understanding of how to use these tools safely.

The requirement extends beyond technology companies and large corporations. Any business using AI tools in routine work may fall within scope, including for tasks such as drafting content, analysing data and screening job applicants.

That broad reach creates a compliance challenge for employers that adopted generative AI tools informally or quickly, often without updating internal controls. Human resources teams are likely to face particular pressure where AI is used in recruitment, employee assessment or other decisions affecting staff.

Crystel Robbins Rynne, Chief Executive Officer of HRLocker, said many employers had yet to realise the obligation was already in force.

"The launch of Ireland's AI Office has created a compliance obligation that many employers have not fully recognised. If your staff are using AI in any part of their work, you now need to show that they are trained to use it safely. This is a significant shift and it's already in force," Robbins Rynne said.

The AI Office is expected to help oversee the law nationally as EU member states put it into operation. For employers, the practical question is less whether they develop AI systems themselves than whether staff are using AI in day-to-day work.

Workplace impact

The issue is likely to be most acute in organisations where AI use has spread through individual teams rather than formal procurement or governance processes. In many offices, workers already use chatbots and other tools to prepare documents, summarise information or support decision-making.

That can leave companies with limited records of who is using which tools, what training they received and what checks are in place before outputs are relied on. It also raises questions about accountability if AI-generated material influences hiring, performance reviews or external communications.

"The real pressure point is that AI adoption has moved faster than organisations were prepared for. Businesses are rolling out tools quickly, but they're not documenting how employees are trained, how outputs are checked, or who is responsible for oversight. That gap is where organisations are now exposed," Robbins Rynne said.

For employers, AI literacy goes beyond technical know-how. It includes understanding when AI can be used, how to verify its outputs, when human review is required and what risks may arise from bias, inaccuracy or inappropriate reliance on automated suggestions.

That matters because some of the highest-risk uses can arise in ordinary office processes rather than specialist engineering settings. Recruitment, data analysis and written communications can all affect people or business decisions, even when the underlying tool appears simple to use.

"A common myth is that AI literacy is a technical skill or something only IT teams need to worry about. In reality, most AI risk sits in everyday decisions such as drafting content, analysing data or screening CVs. If employees are using AI to influence outcomes, employers have a responsibility to make sure everyone understands how to use it properly," Robbins Rynne said.

Basic controls

For many organisations, the first step will be to identify where AI is already in use. From there, employers may need written policies, staff training records and approval processes for higher-risk uses, especially those involving employment decisions.

Robbins Rynne said compliance did not necessarily require complex technical programmes. Instead, she pointed to basic internal measures that employers could document and apply consistently across teams.

"There's a misconception that AI literacy means expensive certification or specialist technical skills. This isn't the case. Irish employers need three things: a clear policy, practical training on how to use and verify AI outputs, and evidence that staff have completed it. If AI is involved in recruitment or employee decisions, you need stronger controls and documented human oversight. These are the basics the AI Office will expect to see," Robbins Rynne said.

The emergence of these rules adds a new layer of governance work for HR, legal and operations teams at a time when many businesses are still experimenting with AI tools. It also suggests regulators will look not only at the design of AI systems, but at how ordinary workplaces deploy them.

"The benefit of acting now is simple. You stay compliant and you build confidence. When employees understand how to use AI safely, they make better decisions, avoid errors and get more value from the tools. Organisations also see stronger governance, cleaner audit trails and fewer risks in recruitment or performance decisions. AI literacy is not a burden. It is the foundation for using AI well and it gives employers a safer, more productive and more accountable way to adopt new technology," Robbins Rynne said.