‘AI does not change an employer’s underlying legal obligations’
AI is rapidly reshaping recruitment in New Zealand, but employers remain legally responsible for errors, bias and misleading information generated by the technology, two experts warn.
Automation is increasingly embedded across the hiring lifecycle, from drafting job ads to shortlisting candidates and running initial interviews, raising complex legal questions for HR leaders.
Simpson Grierson’s employment partner Rachael Judge and AI & technology senior associate Michelle Dunlop say AI is now being deployed to scan and rank CVs against role criteria, automate candidate communications and scheduling, assess recorded video interviews, and in some cases predict a candidate’s “fit” based on historic hiring decisions.
That shift is being driven by the pressure to manage high application volumes at speed, while maintaining consistency in how candidates are treated.
Major employers are already public about their use of these tools, they say. Spark New Zealand has described using an AI “smart interviewer” to streamline early‑stage contact‑centre recruitment and shortlisting at scale, while Qantas has outlined its use of an AI‑driven chat interview to score candidate responses and support early‑round screening. These examples underline that AI in recruitment is no longer experimental or niche – it is moving into the mainstream of corporate hiring.
Jack Malpass, chief operating officer at New Zealand-based hiring platform Weirdly, previously said that the traditional levers HR has relied on – CVs, agency networks and experience-led screening – are no longer fit for purpose in a high-volume, AI-saturated market.
Legal liability: AI errors remain the employer’s problem
Judge and Dunlop are clear that the legal risk attached to AI‑assisted recruitment remains with the employer, not the technology provider. “AI does not change an employer’s underlying legal obligations,” they say. If AI tools generate incorrect information and that content is relied on, HR and business leaders cannot point to the system as a defence.
They warn that problematic content can arise at multiple points in the process. AI systems used to draft or refine job ads and offer letters may produce misstated role requirements, exaggerated benefits, inaccurate pay ranges or misleading claims about flexible work or progression. If those statements appear in published material or contractual documents and candidates rely on them, any resulting legal exposure “will sit with the employer,” say the legal experts.
The potential consequences are wide‑ranging. Depending on the circumstances, employers could face discrimination claims, breach of contract claims, or allegations of misleading or deceptive conduct, they say.
Over 7 in 10 (72%) HR leaders recognise reinventing workflows is essential to unlocking AI-driven productivity, according to a previous report.
Bias, discrimination and the regulatory signal
One of AI’s strongest selling points in recruitment is consistency, but Judge and Dunlop caution that consistency is not synonymous with fairness. Where a model is trained on past hiring data that favoured particular schools, career paths, genders or gap‑free CVs, the system may faithfully reproduce those patterns at scale. “The model will replicate that at scale,” they note.
That has direct legal implications. If an AI‑driven screening process disproportionately filters out candidates because of protected characteristics – such as sex, family status or race – the resulting pattern can amount to unlawful discrimination. Judge and Dunlop say that in those circumstances, employers may be held liable for the actions of a discriminatory AI tool used in their recruitment, even if no individual intended to discriminate.
International regulatory developments are reinforcing these concerns. Under the European Union’s AI Act, tools used for recruitment are considered “high risk”, triggering stricter obligations around transparency, human oversight and continuous monitoring of performance and impact. While New Zealand has not yet adopted an equivalent AI‑specific regime, the EU framework is widely regarded as a benchmark for responsible use, and offers a clear signal of where global expectations on employers are heading.
HR capability, governance and the ‘human in the loop’
According to Judge and Dunlop, HR capability in using AI is highly variable across New Zealand organisations. “There is a spectrum,” they say. “Some of our HR clients are at the very start of their AI journey, whereas others are very proficient and have adopted it as part of their day-to-day work.” Those using AI most effectively tend to have “good governance and clearly communicated guidelines for people around the use of their AI tools.”
They recommend employers focus on education for HR teams “on the benefits and shortcomings of AI”. That includes improving AI literacy, providing clear guidance on when AI can and cannot be used, and offering continuous learning as the technology evolves. Employers should also ensure they can explain, in plain language, “what the tool does and does not do, what inputs it relies on, what it is optimising for, and where human decision-makers intervene”.
Despite the efficiency gains, Judge and Dunlop emphasise that AI should not be treated as the final decision‑maker in recruitment. “When it comes to recruitment, the ultimate decision-making should not rest with AI,” they say. Even if an AI system appears adept at identifying the strongest candidate “on paper”, it “cannot (yet) replace human decision making” about whether someone is the right fit for a specific role and organisation.
They argue for a blended model in which automation handles high‑volume triage, but human‑led interviews and assessments remain central for roles that have material impact on the business.
AI will continue to reshape the way we work, but performance isn’t delivered by technology alone. Human connection remains a decisive factor in how effectively organisations learn, adapt and execute. By intentionally strengthening connection, safeguarding skills and normalising thoughtful trust in AI, HR leaders can protect culture and performance together, according to a previous report.
