AI in Hiring and Management: The Fair Work Risk (2026)

The moment software decides who gets an interview, who gets rostered off, or who gets managed out, your business owns the outcome — not the vendor who sold you the tool. And under the Fair Work Act's general protections, if that decision brushes against a protected attribute, the law flips the burden onto you: you have to prove the algorithm didn't discriminate. Most Australian employers adopting AI in hiring and management haven't priced that reverse onus in.

This is no longer an enterprise-only story. Per the federal government's AI Adoption Tracker, 60 per cent of small businesses with 5 to 19 staff now use AI in some form, and Indeed Hiring Lab reports the share of Australian employers advertising AI-related roles rose from 5.8 to 8.5 per cent in the single year to February 2026. The tools spread faster than the guardrails. The guardrails have now arrived.

In 2026 the Australian Human Rights Commission published an AI and recruitment compliance checklist — the clearest signal yet that regulators expect employers, not software vendors, to answer for automated hiring decisions. The checklist is built on existing anti-discrimination law: an AI screening tool that produces biased outcomes can breach the Sex Discrimination Act or the Racial Discrimination Act, and intent is not required for a finding. A model doesn't have to be told to discriminate to do it.

The subtle trap is proxy discrimination. The AHRC warns that an algorithm can infer a protected attribute it was never given — a postcode standing in for ethnicity, a career gap standing in for gender or disability. Screening tools trained on your past hires quietly learn to reproduce your past hires. If that history skews one way, the tool inherits the skew and repeats it at a scale no human recruiter ever could.

Want to know what this means for your business? We'll map it in one call.

This is where the Fair Work Act bites. General protections claims cover adverse action taken for a prohibited reason — and refusing to employ someone, altering their position, or dismissing them all count. When an applicant alleges an AI tool knocked them back for a discriminatory reason, section 361 of the Act presumes the reason was unlawful unless the employer proves otherwise. As employment specialists at Ius Laboris note, a business relying exclusively on automated decision-making would find it difficult to overcome that reverse burden of proof. You cannot cross-examine a model that can't explain itself.

The exposure doesn't stop at the front door. Most employees under a modern award or enterprise agreement have consultation rights when a major technological change significantly affects them — broad enough, on current legal analysis, to capture AI rostering and performance systems. In New South Wales, workers' compensation reforms now give unions entry rights to inspect 'digital work systems', explicitly tying work health and safety to algorithmic surveillance and decision-making. All of this is unfolding while the Fair Work Commission is already straining: its workload is up more than 70 per cent in three years, the same AI surge now flooding the tribunal with self-lodged claims.

If you're not certain where automated decisions already touch people in your business, start by mapping your own exposure before a claim maps it for you.

The fix is not to abandon AI — it's to keep a human on the hook. The AHRC's own framing is blunt: automate the admin, not the accountability. In practice that means four things. Assign a named person to review every AI-driven rejection before it goes out, so a human signs off on the decision. Choose vendors that publish bias-testing results validated on Australian cohorts, not overseas data. Build an AI register — a plain list of every tool that touches a hiring, rostering or performance decision, including the resume-screening and lead-scoring features already baked into your HR and CRM platforms. And prepare to disclose: from 10 December 2026, businesses covered by the Privacy Act will have to tell people when automated decisions significantly affect them, and whether to grant a job interview is squarely in scope.

None of this makes AI in hiring off-limits. Used with oversight, it screens faster and rosters smarter than any spreadsheet. The businesses that get burned will be the ones that let the tool make the call and kept no record of why.

Key takeaways

Common questions

Related notes

Not sure where AI already makes decisions in your business?

We map where automated tools touch your hiring, rostering and people decisions — and where the liability sits with them. Book a call and we'll walk through it.