Recent news reports highlight a growing anxiety among workers in their 40s, 50s and 60s who cannot progress past the application stage despite decades of experience with some now “botoxing” their CVs to remove references to age. Employers are also turning to AI tools for sourcing candidates and screening CVs without human oversight which can lead to embedding or amplifying existing biases.
Under the Employment Equality Acts, discrimination arises where a person is treated less favourably on one of nine protected grounds: gender, civil status, family status, sexual orientation, religion, age, disability, race or membership of the Traveller community. The Employment Equality Acts cover access to employment, so advertising, shortlisting, interviews and selection are all in scope. Irish HR practitioners will remember one of the earliest access to employment WRC decisions, where a Ryanair job advertisement seeking a "young and dynamic professional" was found to discriminate on the ground of age.
Not every unsuccessful candidate has a claim; however, if a candidate establishes facts from which discrimination may be presumed, the burden shifts to the employer, who must prove otherwise. The key to a successful defence is whether the employer can show, with documents, that the process and outcome rested on lawful, job-related reasons.
How informal comments can become evidence
Most employers know interview questions should not stray into protected characteristics. The harder risk is the well-meaning remark about family life, age, retirement plans, accent or “fit”. Comments about being “energetic”, “too senior”, “near retirement” or “a good cultural fit” can later be relied on as evidence that something other than the needs of the role shaped the decision.
In Emily Williams v Board of Management of St Tola’s National School (ADJ-00055461), Ms Williams, a fixed-term teacher on maternity leave, interviewed for a further fixed-term post. At the end of the interview, the Principal congratulated her and told her to “enjoy every moment at home with the baby”.
After the interview, Ms Williams learned she had been unsuccessful. The role was awarded to had gone to the teacher who covered her maternity cover. She also learned that a less experienced colleague had been given a Contract of Indefinite Duration (“CID”) just as her maternity leave began.
The school said its policy was to award a CID to the highest-ranked candidate from a previous panel, but could not point to any written policy or precedent.
Ms Williams was never told about the CID or given a chance to interview for it. Together with the Principal’s comment, that raised a presumption of discrimination which the school failed to rebut. The WRC awarded her €85,000.
How neutral criteria can cause unlawful disadvantage
Noel O’Connell v National Council for Special Education (ADJ-00042837) shows the risk is not confined to interview comments. Mr O’Connell, who is deaf and a native Irish Sign Language (“ISL”) user, was not shortlisted for an ISL advisor role because he lacked a formal academic ISL qualification.
He argued internally that the requirement was discriminatory as deaf people are far less likely to hold such a qualification in their native language. The decision maker agreed but by then, the recruitment process had closed and no remedy was offered.
The WRC found the qualification requirement amounted to indirect discrimination and awarded €40,000, three times the €13,000 statutory cap for job applicants. To go beyond the statutory limit, the Adjudication Officer relied on Article 17 of Directive 2000/78/EC, which requires sanctions for discrimination to be effective, proportionate and dissuasive.
The EU AI Act: what it means for recruitment in Ireland
Employers using AI in recruitment must also factor in the EU Artificial Intelligence Act (Regulation (EU) 2024/1689) (the “AI Act”). It applies directly in Ireland and is supplemented by the domestic Regulation of Artificial Intelligence Act 2026, which became law on 21 July 2026. It sits alongside the Employment Equality Acts and the GDPR rather than replacing them.
The AI Act classifies AI systems used for recruitment or selection as “high-risk”, catching many CV-screening, ranking and video-interview tools used by employers.
The WRC has also been designated as a Market Surveillance Authority in relation to certain prohibited AI practices in the workplace, including systems that infer the emotions of workers or job applicants, save in limited permitted circumstances.
A candidate who believes an AI tool screened them out on a protected ground can bring a claim under the Employment Equality Acts and also complain about the use of AI to the WRC in its supervisory role.
Human in the loop
Clicking “approve” on an algorithm’s shortlist is not human oversight. The overseer must understand the tool’s capabilities and limitations, guard against “automation bias” (over-reliance on machine outputs) and have authority to depart from its recommendation. Employers should be able to show who reviewed the output, their considerations and why they reached the final decision.
Bias and indirect discrimination
From 2 December 2027, providers of high-risk systems must take steps to detect, prevent and mitigate possible bias. AI systems trained on historical hiring data may favour candidates resembling past hires, and proxies such as graduation dates, career gaps or postcode can disproportionately exclude older candidates, women or particular racial groups. The WRC is unlikely to accept “the algorithm decided” as a defence.
How should employers manage the risk?
The aim is not impersonal interviews but a job-focused process you can explain with evidence. We recommend that employers:
- Prepare questions and scoring criteria in advance, tied directly to the job description and person specification;
- Ask all candidates the same or similar questions and avoid follow-ups that drift into personal life, family status, health, age or other protected grounds. If a candidate volunteers something personal, acknowledge it briefly and return to the requirements of the role;
- Take contemporaneous notes of relevant skills, experience and answers, not personal impressions or comments that could suggest bias;
- Consider reasonable accommodation for candidates with a disability, including adjustments to interview format, location or assessment;
- Agree who will communicate outcomes and feedback, and how records will be retained;
- Carry out due diligence on any AI tools used to source, screen or shortlist candidates, and ensure users are appropriately trained; and
- Keep a human in the loop: AI should support recruitment decisions, not make them.
Conclusion
The WRC decisions show that recruitment claims rarely turn on a single dramatic error. Liability builds from small missteps: a throwaway comment, an unquestioned requirement or an unrecorded decision, and the cost can be significant. As the AI Act brings oversight and transparency obligations into force, employers must explain what their algorithms did as well as their interviewers. A structured, job-focused process with appropriate human decision makers and a clear paper trail will put employers in a better position to defend claims, make better hires and reach a wider talent pool, including the experienced candidates who increasingly feel shut out.
