A BBC report published on 25 August 2026 emphasised that AI-generated complaints are placing a growing and disproportionate burden on schools and local authorities. What it did not address is what happens when that same dynamic moves from a complaints inbox into litigation.

AI in litigation: What we are seeing

AI tools can produce lengthy, structurally complex legal documents quickly. In litigation brought by unrepresented claimants, this is generating claims that look formidable on paper but frequently fail to establish fundamental legal principles: a causal link between the alleged breach and the damage, an accurate application of all strands of the relevant legal test, or any coherent connection between the facts pleaded and the remedy sought. AI can also make errors and misquote legislation, applying legal tests to the incorrect party or advancing causes of action that do not exist in English law.

In a recent case we became aware that an unrepresented claimant used a generative AI programme to prepare a schedule of loss. The output placed the claim at over £150,000. The figure was entirely unsupported by evidence and created such a wide gulf between the parties that negotiated settlement was not possible. The claimant was ultimately awarded less than a tenth of the amount claimed and none of the special damages claimed were recovered. The costs of proceedings for both sides far exceeded what a reasonable early settlement would have achieved.

In KDY v Langham & Others [2026] EWHC 2068 (KB), a High Court judgment handed down on 7 August 2026 in which Browne Jacobson acted for the defendant schools and local authority, the Particulars of Claim ran to 300 pages and 1,009 paragraphs, supplemented by a judicial roadmap, a causation map and two further matrix documents. The entire claim was struck out. The structural features of that claim are consistent with a pattern we are seeing with increasing frequency, even where use of AI has not been admitted.

We are also increasingly seeing disproportionate and onerous disclosure requests, including requests for metadata and electronic disclosure that bear no relationship to the issues actually in dispute - another recognised feature of AI-assisted claims.

Practical steps for schools and local authorities

• Take early legal advice: An AI-assisted claim can look formidable on paper. Early advice will quickly identify structural weaknesses, misapplied law and unsustainable claims for damages. Do not let volume lead you to overestimate the strength of a claim.
• Consider an early strike out application: Where a claim is fundamentally misconceived in its pleading or legal basis, an early application to strike the claim out is a proportionate and effective response. The case of KDY demonstrates that a well-prepared application can dispose of an entire claim at a single hearing.
• Challenge disproportionate disclosure requests: Onerous requests that are not proportionate to the issues in dispute should be challenged promptly, by agreement or by application.
• Support your staff: Where individual employees are named personally as defendants, a tactic we are seeing used with increasing frequency by unrepresented claimants utilising AI, confirm your indemnity position immediately and put pastoral support in place from the outset.

As the BBC noted, AI does not push back. It will not tell a claimant their case is weak, that their schedule of loss is unsustainable, or that they face a costs risk at trial. That is not a reason to take these claims lightly. It is a reason to take early expert advice and to respond decisively.

Browne Jacobson, the Confederation of School Trusts (CST) and PLMR are running a webinar on 17 September to help schools address the growing challenge of AI-generated complaints.