Victoria’s AI Hiring Pledge Is a Preview, Not an Outlier

“We cannot allow discrimination to sneak back into the workplace via AI.” That is Victorian Premier Jacinta Allan, announcing on Monday that her government will regulate how businesses use AI in recruitment and workplace surveillance if Labor wins a historic fourth term in November’s state election. Nothing has passed. No bill is before parliament. This is a policy pledge, tied to a vote still months away.

Treat it as a preview anyway. Whatever happens in November, the standard Victoria is proposing, explainable, auditable, human-governed AI in hiring, is where every jurisdiction regulating this space is heading. The lesson for TA is not to wait and see. It is to already be able to answer the questions this pledge is built to force.

A Pledge, Not Yet a Law

Strip away the election framing and the substance is specific. Labor proposes amending the Equal Opportunity Act to stop AI discriminating by race, gender, sexuality, religion or other personal characteristics during recruitment. The Victorian government’s own release, reported by HCAMag, names the mechanism it is worried about: AI resume-ranking tools that can, in its words, “systematically disadvantage women and people from diverse backgrounds.” It describes bias in AI-driven hiring, promotion and pay as one of the most consequential and least visible harms of the AI transition.

That framing matters more than the election timing. The proposal follows fears of AI bias in recruitment and prior research suggesting AI-driven hiring systems could end up discriminating against candidates.

The Line Between AI That Decides and AI That Shows Its Work

Read the target closely: the government names AI resume-ranking tools specifically, and describes the amendment as targeting discrimination in recruitment generally. Our read is that the practical target is black-box resume ranking and screening, systems that produce a decision, or a score, with no visible reasoning a candidate or a regulator could interrogate. A tool where a human reviewer sees the evidence behind every recommendation and stays accountable for the call sits on the other side of that line entirely.

That is a distinction TA leaders should draw for themselves already, before any legislature draws it for them. AI that decides for you, opaquely, is a different category of risk than AI that surfaces evidence and leaves a human accountable for the judgment. Victoria’s proposed law is effectively legislating a line that good hiring practice should already respect.

The Transparency Bar HR Tech Needs to Clear

The surveillance provisions extend the same logic to monitoring. The proposal would stop AI tracking employee movements and physical attributes, including bathroom breaks, a limp, or pregnancy, without a specific stated purpose. Biometric data could only be collected where no less intrusive option exists, and could not be used to read emotions without a legitimate reason. Employers would have to disclose monitoring to staff and consult them before introducing or changing surveillance technology, with covert surveillance restricted to suspected unlawful activity.

Industrial Relations Minister Jaclyn Symes framed the reasoning plainly: “Every worker has the right to know when, why and how they’re being watched on the job, these laws make sure of it.” She added, “These laws put workers back in control of their own information.” Disclosure, consultation, a stated purpose for every monitoring tool in use: in our view, many HR tech stacks would struggle to meet that bar today.

Victoria Is Not Acting Alone

The proposal did not appear out of nowhere. It follows an inquiry by the Economy and Infrastructure Committee last year that found regulation lagging behind the growth of workplace surveillance. The Finance Sector Union welcomed the package as “a huge win for workers,” noting that its AI campaign and submission to Victoria’s Workplace Surveillance Inquiry had pushed for this for years. The FSU’s own line is blunt: “AI should improve work, not monitor bathroom breaks, make biased decisions, or replace human judgement.”

A parliamentary inquiry finding a gap, a union campaigning on it for years, and a government now proposing to legislate in response is not the shape of an isolated state initiative. We think this is the shape of a standard hardening in one jurisdiction, the kind of pattern that tends to get picked up elsewhere.

What You Should Already Be Able to Answer

Here is the sequence problem. If you are only building explainability and audit trails because a law forces you to, you have already conceded the harm happened first and the fix came second. In our work with hiring teams, the leaders who are ahead of this are not the ones tracking the Victorian election. They are the ones who can already answer three questions for every AI-influenced hiring decision in their pipeline. Who is accountable for it? What evidence backs it? And could a candidate ask “why was I ranked this way” and get a real answer rather than a shrug?

If you cannot answer those three questions today, the gap Victoria’s inquiry found in workplace surveillance exists in your hiring stack too. Legislation forcing you to close it is the slow path. Closing it because it is the right way to run a hiring process is the fast one, and it is available to you regardless of what happens in November.

Original reporting: hcamag.com.