A million applicants, zero guardrails
AI hiring is not arriving in New Zealand. It arrived years ago and kept scaling while nobody wrote the rules. One unnamed New Zealand company screened more than 1,000 candidates by AI in a single month, and that is just one employer. Woolworths has been running Sapia AI since 2020, and more than a million job applicants across Australia and New Zealand have now passed through that single platform. McDonald’s uses AI screening too. 87% of large New Zealand organisations now run AI in operational environments.
The timing amplifies the stakes. Online job advertisements grew 11.8% in the year to March 2026, the third consecutive quarter of annual growth after 11 straight quarters of decline. More employers are hiring, more applications are flowing, and AI is increasingly the gatekeeper.
The efficiency case is hard to argue with
The business logic is compelling. AI-enhanced recruitment can cut a four-week hiring process by 75% and reduce costs by 90%, with CV sifting dropping from 23 hours to five minutes. The cost barrier has collapsed too: 75% of NZ organisations report AI setup costs under $5,000, down from a third previously spending over $50,000.
For a country where labour productivity growth has averaged just 0.5% annually over the past decade, below the OECD average of 0.9%, tools that compress hiring timelines and free HR staff for higher-value work look like a rational response. Sapia AI reports a nine-out-of-ten satisfaction rating across 3.5 million interviews conducted in the past year. Woolworths says a human reviewer follows every initial AI assessment.
None of that is the problem.
Personality assessments from a chatbot
The problem is what happens when the efficiency gains outrun the quality controls. A Kapiti mother reported that Woolworths’ AI told her 16-year-old son he would struggle with distractions and didn’t like to try new things, based on a short text conversation. The NZ Herald documented further applicant experiences involving what it described as “ridiculous assumptions” from a tool making personality judgements it had no basis to make.
Unite Union assistant national secretary Gerard Hehir put the structural concern plainly: “Time and time again over recent years we have seen that of course the processes themselves often reflect the biases of those that wrote them and designed them.” Rather than removing bias, he said, AI tools “reinforce or even amplify the bias.”
Project Employ, which trains neurodiverse New Zealanders for work, raised concerns that algorithmic screening creates structural barriers for people outside conventional norms. Youth unemployment rose to 14.4% in Q1 2026 from 13.3% previously, and that cohort is disproportionately exposed to keyword-matching systems that penalise sparse work histories and non-standard formatting.
AI is now screening AI
The system has produced a self-defeating loop. Sarah Wrightson, who runs a CV-writing business in Te Awamutu, reports her customer base is up 50%, driven by candidates paying to optimise their documents for algorithms. Her concern is direct: “Good candidates are being knocked out of the chance to get the job because of technical reasons, not capability.”
Dr Andrew Lensen, AI researcher at Victoria University, described “a perverse situation where we have people who write lots of applications with AI and then we have employers who are using AI to screen applications”. The employer’s original problem, finding the best candidate, is not solved by this arrangement. It is obscured by it. Signal quality degrades on both ends: strong candidates get filtered out for formatting while AI-polished applications from weaker candidates sail through.
Five questions every employer should answer now
New Zealand has no specific legislation governing AI use in hiring decisions. The Human Rights Act and Privacy Act offer some protection but were not designed for algorithmic personality assessment. There is no right of explanation for rejected candidates, no mandatory disclosure, and no audit requirement. The Algorithm Charter for Aotearoa is voluntary for private employers. The EU, by contrast, classified recruitment tools as high-risk under its AI Act, triggering mandatory transparency and bias auditing.
BusinessDesk’s analysis identified five questions every employer running AI hiring tools should be able to answer: Are candidates told AI is screening them? Is there a human review pathway for rejections? Has the vendor provided bias audit evidence? Is the AI limited to objective criteria or assessing subjective qualities? Can the employer explain how a hiring decision was made?
If the answer to any of those is no, the employer is accumulating legal and talent risk with every application it processes. In 2025, Treasury’s analysis of AI’s economic impact found no evidence of aggregate productivity growth from AI yet, despite positive firm-level impacts. The efficiency gains are real, but so is the gap between what these tools promise and what they can actually be held accountable for. Woolworths was early and visible. The complaints it attracted will become routine across sectors as adoption spreads, and most employers running these tools have not thought past the cost savings.
Sources
- RNZ: Jobseekers and advocates disturbed as companies screen applications with AI (2026-03-27)
- NZ Herald: More complaints arise about Woolworths’ use of AI personality analysis in job interviews (2026-04-03)
- RNZ: AI causing headaches for both job hunters and recruiters (2026-03-27)
- B2B News: AI job interviews NZ – 1,000 screened monthly with no rules (2026-04-08)
- BusinessDesk: AI-assisted hiring – where businesses win and where they get caught out (2026-04-03)
- Treasury: Update on the Impact of Artificial Intelligence – An Economic Analysis (2025-08-05)