A deputy High Court judge sitting in the Employment Appeal Tribunal (EAT), Marcus Pilgerstorfer KC, issued a formal warning on 18 September 2026 that litigants in person (LiPs, meaning individuals representing themselves without a lawyer) who file AI-generated documents without adequate checking face judicial enquiries and potential sanctions. The warning arose in a case involving Terence Hancox, who filed a 300-page, 132,000-word skeleton argument (a written outline of legal arguments submitted before a hearing) which he said had been created using ChatGPT. The document was filed with an instruction to respondents to identify any inaccuracies by 4pm the following day. The judge found this approach unacceptable. Judge Pilgerstorfer cited existing case law and judicial guidance to establish that, while AI use is not generally prohibited for tasks such as preparing skeleton arguments, all litigants including LiPs must take personal responsibility for documents submitted. Minimum requirements include: compliance with procedural rules; thorough accuracy checks for factual, evidential, and legal points including case citations; no misleading of the court or tribunal; and relevance and conciseness, avoiding undue repetition. The judge noted a significant increase in AI tool use before courts and tribunals, with notices of appeal, skeleton arguments, and correspondence all affected. The judge found one ground on which Hancox could appeal and issued case management directions to bring the matter to a hearing-ready state efficiently, warning that a document of this kind would in future be rejected.
Why this matters
This ruling establishes a clear and publicly available standard: the personal responsibility and accuracy duties that apply to legally qualified representatives under existing case law extend equally to unrepresented litigants using AI tools. The EAT's warning has practical significance beyond this case because AI use in litigation preparation is rising sharply, and courts now have a concrete set of minimum obligations to point to when considering sanctions. For law firms, the ruling reinforces that supervising AI outputs is a professional and court-compliance obligation, not merely good practice. The volume and incoherence of the 300-page filing also illustrates a specific risk: generative AI tools produce text at scale, and length without focus is itself a form of procedural non-compliance.
On the Ground
Disputes and litigation practices will use this ruling immediately to update client-facing guidance on AI use in proceedings, and to brief fee earners on the standards courts are applying to AI-generated documents. Risk and compliance teams at law firms will be reviewing their AI governance policies to ensure that internal controls on checking AI-generated court documents are documented and enforced. A trainee assisting on a litigated matter would be expected to conduct a line-by-line accuracy check of any AI-assisted skeleton argument, verify all case citations against primary sources, and prepare a record confirming the checking process, precisely the steps the judge described as minimum requirements.
Interview prep
Question you might get
“What obligations does this EAT ruling impose on litigants using AI to prepare court documents, and what does it mean for law firms advising clients who want to use AI in litigation preparation?”
Sign up free to see the full answer
A model answer you can lift into an interview — how to frame this story for a partner.
Sign up freeMy notes
saved