Job-related AI systems classified as high-risk
Employers in Ireland that use artificial intelligence (AI) for recruitment, workforce management or employee evaluation should prepare for increased regulatory oversight, MHC lawyers have said.
The Workplace Relations Commission (WRC) has been designated as a market surveillance authority (MSA) under the EU AI Act and will supervise certain high-risk and prohibited AI systems used in employment.
The WRC has published a note explaining its new role and outlining what employers should consider when implementing AI in the workplace.
Ireland is adopting a distributed regulatory model for the EU AI Act, the MHC briefing points out.
Rather than creating a single regulator responsible for all AI, existing regulatory bodies will share responsibility.
Under the 2025 regulations, 15 bodies have been designated as MSAs, including the WRC.
An Irish AI Office has also been established to coordinate the national framework and perform certain central functions.
Certain employment-related AI systems are classified as high-risk under the AI Act.
These include systems used for recruitment and selection, such as advertising vacancies, screening applications and evaluating candidates.
They also include systems used to make decisions affecting employment terms, allocate tasks, or monitor and evaluate employee performance.
Employers using these systems are generally regarded as “deployers” and must comply with specific obligations.
Under article 26 of the AI Act, employers must use high-risk AI systems in accordance with their instructions, ensure appropriate human oversight, monitor their operation and report certain incidents.
Employers must also inform workers’ representatives and affected workers before putting certain high-risk AI systems into use. The WRC will supervise employers in fulfilling these obligations.
Provider, not deployer
An employer may also become a “provider” rather than simply a deployer in certain circumstances.
This can occur where an employer develops an AI system itself, has one developed on its behalf and places it into service under its own name or trademark, or substantially modifies or rebrands a third-party system.
This distinction is important because providers have additional responsibilities under the AI Act.
The WRC will also supervise prohibited AI systems involving the inference of workers’ emotions from biometric data.
For example, an AI system that attempts to determine whether an employee is angry based on biometric information could fall within this prohibition.
AI use may also become relevant to existing employment complaints.
Employees can already bring complaints to the WRC concerning matters such as employment rights, equality, discrimination, dismissal and penalisation.
Where AI has contributed to an employment decision, the way in which it was used may form part of the WRC’s consideration of the complaint.
Depending on the issue, other regulators may also become involved. For example, the Data Protection Commission could have a role where personal-data concerns arise.
Human oversight
The WRC has indicated that employers should understand what their AI systems do, what data they use, how they manage risk and how human oversight operates, the MHC note states.
Employers should also assess whether AI systems could produce unfair or discriminatory outcomes and provide individuals with appropriate information where required.
Practical steps include mapping AI use across recruitment and HR, establishing human oversight and accountability, assessing equality, data-protection and employment-law risks, training relevant staff, identifying high-risk or prohibited systems, and reviewing supplier contracts and system documentation.
Employers should therefore begin preparing now rather than waiting for the regulatory framework to mature, MHC said.
From December 2027, relevant high-risk AI systems will also be subject to provider registration requirements in a publicly available database.
Early implementation of transparency, human oversight and accountability measures will help employers manage the emerging regulatory requirements.
Discriminatory outcomes
Employers should also assess whether AI systems could produce unfair or discriminatory outcomes and provide individuals with appropriate information where required.
Practical steps include mapping AI use across recruitment and HR, establishing human oversight and accountability, assessing equality, data-protection and employment-law risks, training relevant staff, identifying high-risk or prohibited systems, and reviewing supplier contracts and system documentation.
Employers should therefore begin preparing now rather than waiting for the regulatory framework to mature, MHC said.
From December 2027, relevant high-risk AI systems will also be subject to provider registration requirements in a publicly available database.
Early implementation of transparency, human oversight and accountability measures will help employers manage the emerging regulatory requirements.