· 4 min read
I have 14 upcoming AI slop trials. Genuine deterrents are only answer
Employers are stuck in a system being gamed by tools that produce plausible-sounding nonsense and a legal framework that offers almost no disincentive.
From the ZheroTax article collection
The numbers are now undeniable. General protections filings in the Federal Court have grown by 153 per cent over the last three years and the number of self-represented litigants has risen by 200 per cent.
From where I sit, defending employers against these claims every day, the picture is stark. Almost every second matter I handle now involves a self-represented litigant armed with ChatGPT or Claude as their legal adviser.
I have watched judges walk into hearings visibly frustrated, forced to wade through pages of AI-generated material before they can even identify what the claim is about. I have sat across from applicants who have received from their AI “adviser” a completely fictitious legal precedent – in one memorable case, an application drawn from US Federal Court rules from Miami – and self-represented applicants who cannot be reasoned with because the chatbot has told them that their constitutional right to natural justice (not a thing, by the way) means they could win $1 million by persisting long enough.
That is the insidious part. AI does not just help people pursue employment claims, it validates them. It reflects back exactly what the user wants to hear, regardless of actual merits.
The result is confirmation bias on an industrial scale, producing applicants with wildly unrealistic expectations and no motivation to resolve matters sensibly. Cases that should settle in conciliation drag through to trial. I currently have 14 upcoming workplace trials driven largely by this dynamic. Each one costs the defending employer tens of thousands of dollars.
The clients paying those legal fees are not all mega-rich corporations. They include aged care providers, childcare operators, not-for-profits, health providers and schools, already operating on thin margins, defending claims from employees whose terminations were substantiated on serious grounds, including misconduct, theft or assault.
Filing a claim has become the automatic step at the end of the employment life cycle. Sue your employer on the way out. Why not? It costs almost nothing to try and the downside risk is negligible.
That is the fundamental problem the government has failed to address with its proposed Fair Work Court. Building a new institution will not reduce the number of claims, it will simply give those claims a new venue. As town planners have discovered, if you build more highways to reduce traffic congestion, you get more cars on the road.
What is actually needed is a raft of genuine deterrents. First, the Fair Work Act must be amended to allow costs to be awarded against an applicant whose claim has failed, or had no reasonable prospect of success, where AI-generated material has been a material contributor to that outcome.
At present, the costs protections that make the Fair Work jurisdiction accessible also make it unworkable. That imbalance must be corrected.
Second, if access to legal advisers is genuinely cost-prohibitive for most individuals, then legal aid should be made available as it is in family and criminal law disputes, permitting individuals to be represented by state-appointed lawyers through to the conclusion of their matter. A qualified human lawyer can counsel a client on the actual merits of their claim objectively, in a way that a chatbot never will.
If the government is serious about managing the cost of this AI wave without simply appointing an army of new judges, it should also consider a more confronting solution: AI-assisted mediation. Deploy algorithmic mediators to assess, negotiate and resolve the claims that AI has generated in the first place. Fight fire with fire.
I believe access to justice is worth defending. But it was never meant to mean unlimited AI output, zero cost, and no accountability. The Fair Work Commission’s own president calls it “unsustainable”.
Workplace disputes focus on the intricacies of human relationships, yet litigation in this space is becoming a synthetic experience of AI-generated, volume-produced, unchecked slop. The damage falls not only on the legal system but on real people bearing the costs, the time, and the toll of defending against it.