There’s no room in court for AI hallucinations
Andrew Stevenson looks at the challenge of non-legal parties using AI
Throughout history, elites have preserved power by curbing or prohibiting access to technology or information. When these restraints have fallen there can be revolutionary change, such as peasants armed with crossbows or the Bible being translated into English. In the hands of laypersons, AI is likely to cause disruption to conventional court procedure in ways that seem hard to resist or control.
Previous barriers have collapsed; access to legislation is now freely available and court websites publish decisions. This allows any literate citizen to frame a legal argument. AI goes much further; it is not merely a research tool, but one which will also undertake that framing and do so competently given the correct prompts. In many ways, AI is a great leveller in that unqualified people now have access to IT that will not only research the law, but easily draft writs, defences and other documents that can be used in litigation.
The prevalence of AI use among lay court users is unclear, and probably impossible to quantify, but it is likely to be very common and increasingly so. The courts do not yet appear to have in place any AI detection or screening system which would, for example, identify fictitious case law and legislation. Instead, The Courts and Tribunals Judiciary AI Guidance for Judicial Office Holders (October 2025) suggests judges apply their instincts and look for clues – Submissions that use US spelling may be a sign of litigants having used AI.
Lawyers are officers of the court and heavily regulated. Laypersons are neither and so difficult to sanction.
The ease with which AI assists is seductive, and appears all-knowing and infallible. This reliance was evident in the Kirkcaldy case Your Home Partners v Kellichan and Hood [2026] SC KDY 34, where chaos ensued. When the court explained the statutory position to the pursuer, the latter disputed it and referred to case law and statute. This could not be located by the court for an excellent reason; it was fictitious.
The sheriff noted the pursuer claimed to have found this material “using online Artificial Intelligence in good faith”. That “the lodging of false legal references has the potential to obstruct justice”. And there was “wasting court time and misleading the court”.
The sheriff considered contempt of court, referring to R (Ayinde) v Haringey LBC (DC) [2025] 1 WLR 5147, where the court opined that “Those who use artificial intelligence to conduct legal research notwithstanding [the] risks have a professional duty therefore to check the accuracy of such research by reference to authoritative sources, before using it in the course of their professional work” Laypersons, however, have no “professional duty”.
Lord Justice Baker noted in D (A Child) Recusal [2025] EWCA Civ 1570: “Litigants in person are in a difficult position putting forward legal arguments. It is entirely understandable that they should resort to artificial intelligence for help. Used properly and responsibly, artificial intelligence can be of assistance to litigants and lawyers when preparing cases.” He warned: “All parties – represented and unrepresented – owe a duty to the court to ensure that cases cited in legal argument are genuine and provide authority for the proposition advanced.”
One problem remains, however; it is a little harsh to forgive party litigants for using generative AI but complain when they then lack the access to traditional sources to verify it.
Andrew Stevenson, Secretary, Scottish Law Agents Society