Britain's employment tribunals are being tested by a tool that costs nothing to use: ChatGPT and Grok. Workers with a workplace dispute no longer need to pay a lawyer to write a legal claim. They can ask a chatbot to do it for free, and thousands are doing exactly that.
The clearest sign is a specific type of urgent request called interim relief, which lets a fired worker ask a tribunal to keep their job or pay flowing while their case is heard. Tribunal presidents Barry Clarke and Susan Walker reported that this request went from about 20 filings a year across the whole system to about 20 filings a month at every single office. In June 2026, they issued new guidance telling judges to apply a stricter test before granting it, since most of these AI-drafted filings run hundreds of pages and are packed with fabricated laws and demands no tribunal would ever grant.
The broader numbers back up the strain. New claims rose 39 percent over the year to March 2026, while the tribunal managed to close only about half of what came in. The unresolved backlog grew 55 percent, reaching 64,000 open cases.
It is about to get worse. The Employment Rights Act, working its way through implementation, removes the cap on unfair dismissal payouts and cuts the waiting period before a worker can sue from two years down to six months. More people will qualify to file, and each successful claim will be worth more, which gives everyone a bigger reason to try.
None of this is unique to Britain. Australia's Fair Work Commission says its workload is up 70 percent over three years, and it now requires anyone filing with AI help to say so. A recent academic study found that lawsuits filed without a lawyer in United States federal courts have nearly doubled since ChatGPT went mainstream, with roughly one in five complaints now containing AI-written text. A federal judge in Minnesota called the trend "an existential threat to the federal courts."
There is a more hopeful data point buried in all this. A study out of Pakistan found that when judges themselves were trained to use an AI assistant, they resolved more cases, faster, delivering a return of nearly 40 dollars for every dollar spent on the program. The difference is who the tool serves. Handed to claimants with no guardrails, AI multiplies paperwork faster than any court can read it. Handed to the people managing the caseload, with proper training, it clears the pile instead of growing it.
For any business, the lesson is blunt. Every dispute, real or invented, now costs more to close because someone has to read it first. Legitimate workers with real grievances wait longer for a hearing, while employers pay lawyers to answer claims that a chatbot generated in minutes. Expect UK-style filtering, fees for weak claims, and mandatory AI-disclosure rules to spread to wherever a similar tribunal or small-claims system exists. Businesses that build a clear, boring, well-documented dismissal process now will spend far less defending it later, since a messy paper trail is exactly what a hundred-page AI-drafted claim is built to exploit.