Generative AI Swamps Australia's Labor Tribunal: Workload Up 70% in Three Years

Australia's Fair Work Commission has announced an urgent review of its processes, facing an unprecedented influx of claims filed with the help of generative AI tools. Over three years, the commission's workload has grown by 70%.

How AI increased the workload

Generative AI has democratized access to legal tools. Previously, filing a claim meant hiring a lawyer or spending hours studying the law on your own. Now it is enough to describe the problem to ChatGPT, and within minutes a legal document is ready.

Tools such as ChatGPT, Claude and other models can quickly generate complaints, applications and other documents. This lowered the barrier to entry: workers can now defend their rights without lawyers.

It sounds good in theory. In practice, the commission received an avalanche of paperwork. The Fair Work Commission is not ready for this volume: staffing has stayed the same, while the number of cases has tripled.

Quality problems

Accessibility and accuracy are pulling in opposite directions. Generative AI often produces documents that are formally correct but wrong in substance. Filings generated by models contain:

• Hallucinations: invented facts and statutes
• Logical errors: contradictions within a single document
• Incorrect formatting: documents that are technically not accepted by the court
• Duplication: the same case filed several times from different angles

Commission staff have to examine each case from scratch and check sources to work out what a person wrote and what the neural network made up. This takes more time, not less.

How the commission will adapt

The Fair Work Commission has announced a review of its processes but has not yet named specific measures. There are several options: tighten the requirements for the form of filings, hire more staff, or deploy AI tools on the commission's side to check documents automatically.

The main challenge: the commission cannot simply shut the door on AI. First, that would be unfair, since not everyone can afford a lawyer. Second, it is technically impossible to control.

What this means

Australia's story shows the real price of democratizing tools through AI. On one hand, workers gained access to justice. On the other, courts faced administrative overload.

The question is not whether AI is good or bad. The question is whether institutions (courts, administrations, universities, hospitals) are ready for tools that change the scale of work by 3-10 times within a few years. Australia is looking for an answer right now. Other countries should watch closely.