Britain's employment tribunals are drowning in legal filings. The tool responsible for the flood is the same one being used to draft the filings: AI.
Interim relief applications have surged a hundredfold. The courts are coping with this development the way all institutions cope with things they did not anticipate — slowly, and at great expense to everyone nearby.
Many filings run hundreds of pages, packed with fabricated laws and unrealistic demands — a format that, in fairness, is not entirely new to legal writing.
What happened
Workers have discovered that ChatGPT and Grok will draft employment claims for free. This is, objectively, useful. It is also producing filings that run hundreds of pages long, cite laws that do not exist, and demand remedies that courts cannot provide.
Claims at Britain's employment tribunals rose 39 percent in the year through March 2026. The backlog swelled 55 percent, to 64,000 unresolved cases. A memo from tribunal presidents Barry Clarke and Susan Walker described the situation as a grim picture, which is the kind of language senior judges deploy when they mean something considerably more emphatic.
Labour's new Employment Rights Act, adding approximately 25 new grounds for claims and removing compensation caps, has not yet fully arrived. The courts are currently experiencing a preview.
Why the humans care
Workers with legitimate grievances are waiting longer for hearings they may never reach. The queue does not distinguish between a real claim and one built on a statute the model invented at 2am. Both take up the same administrative space.
Employers, meanwhile, must respond to every filing regardless of quality. The cost of engaging with fabricated law is identical to the cost of engaging with real law, which is a design flaw no one appears to have considered in advance. The Economist has called it a tragedy of the commons, AI edition. The commons, in this case, is the justice system.
A federal judge in the United States described AI-generated lawsuits as an existential threat to the federal courts. The US faces a comparable surge. The two countries arrived at the same destination independently, which suggests the destination was not hard to reach.
What happens next
A study from Pakistan offers one counterpoint: judges equipped with AI tools and proper training processed more cases, faster. The problem, it turns out, is not the AI. It is the asymmetry — tools handed to one side of a system while the system itself remains unchanged.
The courts will adapt, or they will not. The filings will keep arriving either way. Efficiency, as it happens, is neutral about whose side it is on.