Sheriff MacRitchie rejected the claimant’s argument that, once the tenancy had ended, the arrears constituted a standalone civil debt recoverable in the Sheriff Court, finding instead that rent arrears continue to “arise from” the tenancy regardless of its termination. Applying a wide interpretation of section 71, supported by Parker v Inkersall Investments Ltd and Anderson v Stark, the court confirmed that housing-related disputes, including post-termination arrears claims, should be determined by the FtT and not the Sheriff Court. Accordingly, the claim was dismissed and it was held that the Sheriff Court had neither jurisdiction nor competence to make the order sought.
This decision is however notable for its explicit judicial warning on the misuse of artificial intelligence in legal research, even by lay parties. The claimants relied on purported FtT decisions, statutory provisions and procedural rules that were later found not to exist, all of which they conceded had been generated using online AI tools. The Sheriff in this case accepted that this was done in good faith and, on the basis that the claimant withdrew the references before the discussion, he ultimately declined to initiate contempt proceedings. However, it was only after a careful consideration of whether the conduct risked obstructing the administration of justice.
Importantly, the court made clear that good faith is not a complete answer where AI-generated material is utilised uncritically. Sheriff MacRitchie emphasised that false legal authorities had directly influenced the court’s procedural handling of the case, including the decision to assign a hearing which would otherwise not have been necessary. This unnecessary use of court time was treated as a serious matter, sitting “on a fine line” between recklessness and contempt. The Sheriff expressly endorsed the principle articulated in R (Ayinde) v Haringey LBC (DC) [2025] 1 WLR 5147, paragraph 7 (per Dame Victoria Sharp P) where the court stated:
“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 (to advise clients or before a court, for example)”.
Indeed, the Sheriff extended that expectation beyond legal professionals to lay litigants, given the potential systemic impact on court resources and case management.
The judgment therefore stands as an early authority signalling that Scottish courts will not treat AI-generated legal submissions, whether by solicitors or lay persons, any differently to any other court documents. While the Sheriff stopped short of sanction in this instance, the reasoning makes clear that future litigants may not be afforded the same latitude. The decision implicitly draws a line between AI as an assistive tool and AI as a substitute for legal verification, warning that “reckless” reliance can expose parties to procedural dismissal and potentially, and most dangerously, contempt. In that sense, the case is as much a cautionary lesson on the use of AI in litigation as it is a jurisdictional housing decision.
Artificial intelligence is becoming an increasingly visible feature of day-to-day legal practice. During a recent court appearance in which the author of this article appeared, the Sheriff invited parties to confirm whether artificial intelligence had been used in the preparation of written submissions lodged in process. While no express view was taken on the appropriateness of such use, the fact that the inquiry was made suggests a growing judicial awareness of, and engagement with, the role of AI in legal preparation. Developments in this area will continue to be monitored, with particular attention to their potential implications for court users.
