A stack of paper documents. Image: Niklas Bildhauer / Wikimedia Commons, CC BY-SA 2.0, cropped
A “skeleton argument” is supposed to be short. The UK’s Employment Appeal Tribunal recommends 5 to 15 pages, with a hard limit of 20. So when a man representing himself filed a ChatGPT-written skeleton argument that ran to 300 pages and almost 132,000 words, around 10,000 words longer than Pride and Prejudice, the judge was not impressed, and used the case to set out what everyone using AI in court must do, ICLG reports.
What happened
The case, Hancox v Sutherland and others, is an appeal by Terence Hancox, a former handyman at the National Farmers’ Union who brought whistleblowing and discrimination claims after he was dismissed in 2020. An employment tribunal struck out his claim against four individuals in 2021, finding he had conducted the proceedings in a “scandalous, unreasonable and vexatious” manner, including through a social media campaign on LinkedIn.
Hancox appealed. Two days before a July hearing, he filed his skeleton argument, which he later told the judge he had created with ChatGPT because he had to prepare it quickly. In a covering email, he admitted he didn’t expect the judge to read every page, and asked the other side to find any “factual inaccuracy,” “incorrect quotation” or “mistaken date or time” in it by 4pm the next day.
‘Entirely unacceptable’
In a judgment handed down on September 17, Deputy High Court Judge Marcus Pilgerstorfer KC didn’t hold back:
The document filed was entirely unacceptable. It did not come close to complying with the requirements of the PD.
Marcus Pilgerstorfer KC, Deputy Judge of the High Court
As well as being far too long, the document was split into sections with their own paragraph numbering, didn’t use a consistent 12-point font, and referred to documents outside the agreed bundle. The judge said it “served to obscure rather than to illuminate.” At the hearing, Hancox told the judge that much of the document “lacked credibility,” and in the end nobody relied on it.
The judge was just as critical of the attempt to make the other side check it:
It was for the Appellant to take responsibility for the contents of his own written argument. It is not acceptable for a litigant to attempt to place that burden on their opponent, far less to do so over an unreasonable timescale.
Marcus Pilgerstorfer KC
The new rules for using AI in court
The judge was clear that there’s no ban on using AI, and that it can help people who can’t afford a lawyer. But he said the same principles of personal responsibility and accuracy that apply to lawyers apply to everyone. Anyone using AI should make sure that every document they submit, at a minimum:
- complies with the procedural rules, such as page limits for skeleton arguments
- has been checked for accuracy as thoroughly as they reasonably can, so that the facts, evidence and legal references are correct and the tribunal isn’t misled
- contains only relevant points, focused on the strongest arguments, easy to understand and without undue repetition
Simply submitting the product of generative AI to a court or tribunal, or placing the onus to check a document onto an opponent, is not acceptable.
Marcus Pilgerstorfer KC
He warned that where it looks like those checks haven’t been done, “judicial enquiries and potential sanctions are likely to ensue.”
The ruling builds on last year’s landmark case R (Ayinde) v London Borough of Haringey, in which the High Court warned lawyers that tools like ChatGPT “are not capable of conducting reliable legal research” and can invent cases and quotes. This judgment extends the same expectations to people representing themselves.
The twist: he partly won
Despite everything, Hancox didn’t leave empty-handed. Helped by a volunteer barrister from the Employment Law Appeal Advice Scheme, who focused on a single point, he persuaded the judge that it was “reasonably arguable” he hadn’t been given fair notice that the strike-out application would be decided at the 2021 hearing, or a fair chance to prepare. That ground will now go to a full appeal.
His other applications didn’t fare so well. The judge dismissed an attempt to add a further human rights ground, certified a disclosure application as “totally without merit,” and imposed tighter case management rules after what he called an excessive volume of correspondence and applications.
Why it matters
More and more people are turning to chatbots to fight their own legal battles, because lawyers are expensive and AI is free or cheap. This ruling shows the upside and the downside in one case: the AI produced an unusable 300-page document, but a short, focused argument still won the day. The message from the tribunal is simple. AI can help you, but you’re responsible for every word you hand to a judge.
Sources: Hancox v Sutherland and others [2026] EAT 139, ICLG, Law Gazette
Related
