Somewhere in your organisation there’s probably software reading CVs faster than any recruiter could, and I’d bet a fair few executives couldn’t say exactly what it’s scoring people on. The tools got good quickly, and most of us signed up before anyone sorted out who’s accountable when they get it wrong.
So here’s my answer up front. The tech isn’t really the hard part anymore. What HR leaders need is the judgement and governance to sit over the top of it, because Australian law is catching up and “the algorithm did it” won’t wash with a regulator or a candidate who’s been knocked back.
How Many Australian Employers Are Actually Using AI to Hire?
More than you’d think. The Australian HR Institute’s Quarterly Work Outlook for the December quarter of 2025, drawing on more than 600 senior HR people and business leaders, found 49 per cent of employers weren’t using AI anywhere in recruitment or selection, so flip that and just over half were. Remote’s Global Workforce Report 2025, based on 300 Australian organisations, found close to a third had brought a new AI tool into hiring in the previous twelve months.
The same AHRI report found 68 per cent of organisations have formal AI policies, which sounds decent until you notice roughly one in four have nothing at all.
Why Can’t You Just Trust The Software To Be Fair?
Because it learns from us, and we haven’t always been fair. Everyone in HR tech knows Amazon’s story, reported by Reuters in October 2018. Its experimental CV screening engine was trained on a decade of applications that mostly came from men, and it ended up penalising CVs that included the word “women’s” and downgrading graduates of two all-women’s colleges. The team was eventually disbanded.
Closer to home, a 2025 study by legal researcher Natalie Sheard, covered by HRD Australia, warned that AI hiring systems used by Australian employers “may solidify traditional forms of hiring discrimination”, partly because training data can carry past bias and model features can act as stand-ins for protected attributes. HRD also cited AHRI data showing about two in five HR leaders who use AI in recruitment think it discriminates against under-represented groups.
What Does Australian Law Actually Expect Of You?
More than people assume. The Sex, Racial, Disability and Age Discrimination Acts all cover employment, and the Fair Work Act’s adverse action rules include not hiring someone, and I’m not aware of anything in them that lets an employer off because software made the call.
Privacy’s the bit that’s about to change. Come 10 December 2026, thanks to the Privacy and Other Legislation Amendment Act 2024, any organisation covered by the Privacy Act that feeds personal information into automated decisions affecting people’s rights or interests has to own up to it in its privacy policy, spelling out what kinds of information and decisions are involved. The OAIC has a page on the automated decision-making obligation, and when its final guidance came out on 30 September 2026, two things jumped out at me. Having a person look over the output doesn’t get a tool off the hook if what it spits out is a material input to the decision, and it’s you (the organisation using it) that has to disclose, not the vendor who sold it to you.
Could The Rules Get Tougher?
Maybe. Back in February 2025, a House of Representatives committee (Employment, Education and Training) put out its Future of Work report, and the very first thing it recommended was treating AI used for employment purposes as high-risk. It also wanted workers and job applicants told when AI’s being used, and a ban on final decisions with no human oversight, HR decisions especially.
Then the government went the other way, sort of. Its National AI Plan in December 2025 dropped the idea of mandatory guardrails for high-risk AI and said existing, technology-neutral laws would do the job. So for now there’s no hiring-specific AI law, though if I were running HR I’d treat that committee report as a pretty good hint about where expectations are going.
So What Do Hr Leaders Need Beyond The Tools?
Honestly? Mostly governance. I’d start with a bias audit, looking at who’s getting screened out (broken down by group) before a tool goes live and then again every so often, and I’d keep the records, because the day a complaint turns up you’ll want your own evidence, not the vendor’s glossy brochure.
The human oversight needs to be the real thing, too. If your recruiter just clicks “approve” on every shortlist, that’s not oversight, so somebody has to have both the authority and the time to overrule the score when something looks off. I’d also tell candidates when AI’s involved and give them a way to push back on an outcome, which is pretty much what the AI and recruitment compliance checklist from the Australian Human Rights Commission (put together with LexisNexis) asks for.
Now that the privacy duty lands on you and not the vendor, it’s worth grilling them a lot harder about what data the model learnt from and what it’s actually measuring. And please don’t skip the change management side. Sheard’s study recommended training the people who run and oversee these systems, which makes complete sense to me, because a recruiter who knows where a tool falls short won’t trust it blindly.
Where Do You Build Those Skills?
The University of Melbourne’s online Master of Human Resource Management (HRM) covers a fair bit of that ground over 24 months, and it’s AHRI accredited, with subjects including HR Data and Analytics, Law and HRM, and Change and Transformation, which introduces using generative AI to identify change issues.
I suspect the organisations that do well here won’t have the flashiest screening tool. They’ll have someone in HR who can explain, calmly and in plain English, why a candidate didn’t get through, and I reckon that’s a job worth getting good at.
