Artificial intelligence has quietly moved into the parts of employment that used to be handled by managers and HR: screening applicants, scoring performance, building rosters, flagging conduct. When one of those decisions ends up in dispute, it does not land in a special AI court. It lands at the Fair Work Commission, Australia's national workplace relations tribunal, and it is tested against the ordinary requirements of the Fair Work Act 2009 (Cth).
That matters because the Commission's questions do not change when software is involved. Was there a valid reason. Was the process fair. Was the employee consulted where they were entitled to be. An employer that cannot answer those questions is in the same position whether the decision came from a spreadsheet, a supervisor, or a model. This explainer sets out what the instrument requires, where AI enters the picture, and what a risk or compliance operator should put in place.
What the Fair Work Act and the Commission require
The Fair Work Act 2009 (Cth) is Australia's principal national workplace relations statute. The Fair Work Commission is the independent tribunal that administers much of it. It is important to be precise about the Commission's role: it adjudicates disputes between employers and employees. It is not an AI regulator, and it does not licence or certify technology. Its relevance to AI is indirect but real, because AI-assisted decisions become visible through the claims employees bring.
Unfair dismissal and general protections
Two familiar claim types do most of the work here. In an unfair dismissal claim, the Commission considers whether a dismissal was harsh, unjust or unreasonable, which turns on whether there was a valid reason and whether the process was procedurally fair. In a general protections claim, the focus is on whether adverse action was taken against an employee for a prohibited reason, such as the exercise of a workplace right. Both invite scrutiny of how a decision was actually made.
The consultation obligation
Separately, many modern awards and enterprise agreements contain consultation clauses. These typically require an employer to consult employees about major workplace change that is likely to have a significant effect on them. Introducing AI or automation into work processes can be exactly that kind of change, which means the obligation to consult can be triggered before a system goes live, not after it causes a problem.
Where AI comes in
The AI does not create a new legal category. It changes how, and how visibly, decisions get made, and each pathway maps onto an existing obligation.
Recruitment and hiring
AI-assisted screening, ranking and video-assessment tools shape who is shortlisted and who is rejected. Selection decisions can carry discrimination and general protections exposure, and an opaque model that filters candidates on unexamined criteria makes that exposure harder to defend and harder even to detect.
Algorithmic performance management and rostering
Systems that score productivity, allocate shifts or rank workers increasingly feed real employment consequences: reduced hours, missed progression, or the evidentiary basis for a dismissal. If a performance-driven dismissal is challenged, the employer has to stand behind the reason. "The system flagged them" is not a valid reason on its own, and a manager who cannot explain what the system measured will struggle to show the reason was sound.
AI-influenced dismissals
The sharpest risk is a dismissal that leans on an automated output. Before the Commission, "the algorithm decided" is not a defence. The employer must still demonstrate a valid reason grounded in facts and a fair process, including that the employee understood the case against them and had a genuine chance to respond. An automated recommendation the employee never saw and could not contest cuts against procedural fairness.
Always-on monitoring and the right to disconnect
The Fair Work Act now includes a right to disconnect, under which an employee may refuse to monitor, read or respond to contact outside working hours unless that refusal is unreasonable. Always-on AI monitoring and productivity tools sit directly across this right: a system that expects constant availability, or penalises employees for not responding after hours, can collide with a protection the Act now recognises. Workplace AI surveillance also engages separate state and territory surveillance laws, which sit outside the Fair Work framework and carry their own requirements; treat that as a distinct obligation to check, not covered here.
What to do
Keep a human decision-maker accountable. For every consequential employment decision, name the person who owns it. AI can inform the decision; it cannot be the decision-maker of record.
Preserve procedural fairness. Ensure the employee knows the substance of any concern, including anything an automated tool contributed, and has a real opportunity to respond before a decision is finalised.
Consult where you are required to. Check the relevant modern award or enterprise agreement before introducing AI or automation, and follow any consultation clause that applies to major workplace change.
Document the basis for AI-assisted decisions. Record what the tool considered, what a human reviewed, and the reasons the decision-maker relied on, so the reasoning can be reconstructed later.
Respect the right to disconnect. Review monitoring and always-on tooling so it does not, in effect, compel out-of-hours availability, and check state and territory surveillance requirements separately.
Test recruitment tools for fairness. Understand what any screening or ranking model actually optimises for, and keep the criteria explainable to a human.
The pattern across all of this is simple. AI can assist employment decisions, but it cannot absorb accountability for them. The Fair Work Commission will keep asking whether there was a valid reason and a fair process, and the answer has to come from a person who can explain it. If you want a fast, structured read on where AI sits in your own hiring, performance and monitoring practices, the free AIRA Health Check is a practical place to begin.