Blog

AI and the Fair Work Act: Consultation Obligations When Automating

July 2026 · 6 min read · AI Strategy

Two people in consultation across a table with a terracotta speech bubble, a gear and a document nearby
← Back to all posts

Automating a workflow with Claude usually starts as a productivity question: which repetitive tasks can a capable AI assistant take off your team's plate? For Australian businesses there is a second question that often gets skipped, and it can be expensive to miss. When automation changes what a role does, or removes the role entirely, the Fair Work Act 2009 (Cth) may require you to consult affected staff before you act. Get the sequence wrong and a sensible efficiency project can turn into an unfair dismissal claim.

This is not a reason to avoid automation. It is a reason to plan the people side and the technology side together. Here is how the consultation duty works, when an AI project triggers it, and a practical sequence that keeps a rollout on the right side of the line.

Where the consultation duty comes from

The obligation to consult does not sit in one tidy section of the Act. It lives in the consultation term that every modern award and enterprise agreement is required to contain. Section 205 of the Fair Work Act requires enterprise agreements to include a consultation term, and modern awards carry a model term with similar effect. The duty is triggered when an employer makes a definite decision to introduce a major change to production, program, organisation, structure or technology that is likely to have significant effects on employees. Introducing AI to do work that people currently do sits squarely inside 'technology', and often inside 'structure' as well.

'Significant effects' is defined in the model term, and the list is broad:

  • Termination of employment

  • Major changes to the composition, operation or size of the workforce, or to the skills the work requires

  • The elimination or reduction of job opportunities, including chances for promotion or tenure

  • Altered hours of work

  • The need to retrain or redeploy employees

  • The restructuring of jobs

When an AI project crosses the line

Not every automation project triggers consultation. The dividing line is whether the change has a significant effect on employees, not whether new technology is involved. A tool that helps a person do their existing job faster, with the same role and the same headcount, usually will not. A change that removes the tasks a role was built around, alters the skills the job needs, or makes a position redundant usually will.

Two examples make the difference concrete. If a support team uses Claude to draft first-pass replies and the plan is to reduce the team from six people to four, that is a change to the size of the workforce and a likely redundancy, so the consultation duty is live. If a lawyer uses Claude to condense long contracts into shorter briefs while keeping the same role and responsibilities, there is usually no significant effect and no duty. The technology is identical in both cases. The effect on people is what decides it.

Timing is the part businesses most often get wrong. The duty bites once the decision is definite, and consultation has to happen before the change is implemented, not announced after the fact. Rolling out the automation first and telling staff afterwards is the pattern that generates claims.

The cost of skipping consultation

The sharpest consequence shows up in redundancy cases. Under section 389 of the Fair Work Act, a dismissal is only a 'genuine redundancy' if the employer complied with any obligation to consult in the applicable award or enterprise agreement. Skip the consultation and the redundancy is no longer genuine, which removes the employer's main defence and opens the door to an unfair dismissal claim.

That exposure is not trivial. Unfair dismissal compensation is capped at six months of the employee's pay. For a role paid $95,000 a year, that is close to $47,500 before you add legal costs, the management time a hearing consumes, and the possibility of reinstatement. A poorly handled automation of two or three roles can carry six-figure risk, and a single mishandled dismissal can cost more than the first year of savings the project was meant to deliver. There is also a quieter cost: the staff who remain watch how the change was handled, and trust is slow to rebuild.

How Claude fits a compliant rollout

The businesses that automate well in Sydney and across Australia treat the compliance work as part of the project, not a hurdle bolted on at the end. A workable sequence looks like this:

  • Map the affected roles before you build, and write down what each role actually does today

  • Decide honestly whether the change will produce a significant effect, using the model term list as your test

  • If it will, notify affected employees and any representatives once the decision is definite, and put the relevant information in writing

  • Discuss measures to avert or lessen the effects, and genuinely consider what staff raise before finalising anything

  • Look at redeployment and retraining first, since automation usually frees capacity for higher-value work rather than simply cutting it

  • Keep records of every step, because the paper trail is what proves the consultation happened

Claude is useful on both sides of this. It can do the automation, and it can help produce the governance layer around it: a plain-English change summary for staff, a consultation information pack, a redeployment options analysis, and an FAQ that answers the questions people will actually ask. Building that documentation alongside the technical work keeps the rollout defensible and the team informed.

None of this is a reason to slow down on AI. Australian businesses that plan the employment obligations and the automation together move faster, because they are not stopping to untangle a dispute later. If you are weighing an automation project that touches a role, we help scope it so the productivity gains and the Fair Work obligations are planned in the same breath.

Want a second set of eyes before you build? You can book a brainstorm to talk it through.

This article is general information, not legal advice. Consultation obligations depend on the specific award or enterprise agreement that covers your staff, so check the terms that apply to your business or speak with an employment law adviser before you act.

Ready to move from AI pilot to production?

We help mid-market Australian businesses deploy AI automations that actually reach production and deliver measurable ROI.