Scottish judge warns of contempt proceedings after landlord uses AI-invented law

TL;DR: A landlord citing AI-generated legal references in Kirkcaldy Sheriff Court was found to have relied on fabricated legislation and fictitious case law. Sheriff John MacRitchie stopped short of a contempt finding but warned that submitting false references risks obstructing justice. The case is thought to be the first time AI-generated legal hallucinations have been addressed in Scots law.

A property company seeking to recover £5,000 in rent arrears from former tenants told Kirkcaldy Sheriff Court it was relying on the Interest on Debts (Scotland) Act 1985. The problem: that Act does not exist.

Your Home Partners, which was acting without legal representation, also cited tribunal decisions supporting its position. Court staff contacted the relevant tribunals and confirmed those cases were also fictitious. The company later admitted it had used AI to prepare its case.

A “fine line” on contempt

Sheriff John MacRitchie described the company’s approach as reckless but concluded that the partners had not knowingly attempted to interfere with the administration of justice. In Scotland, contempt of court carries penalties of up to two years in prison or an unlimited fine.

In his written judgement, the sheriff drew a distinction between deliberate deception and careless reliance on unverified AI output. “There is a fine line in this instance between whether the claimant and its individual partners, even as lay persons, have shown contempt for the court by not reasonably checking that such references were genuine,” he wrote.

He dismissed the rent arrears claim, ruling that the case should have been heard by a specialist tribunal rather than the sheriff court.

The Kirkcaldy case follows a similar English ruling involving Haringey Law Centre and Haringey Council, where a lawyer faced action for relying on fabricated legal citations. In the US, a New York lawyer was sanctioned in 2023 for submitting ChatGPT-generated case citations that turned out to be invented, a case widely regarded as an early warning.

French legal scholar Damien Charlotin has tracked a sharp increase in AI legal hallucinations across jurisdictions, reporting that incidents rose from roughly two per week before spring 2025 to two or three per day by autumn.

Looking forward

For UK businesses and individuals using AI for legal research, the sheriff’s warning is direct: AI-generated legal references must be independently verified before being submitted to any court or tribunal. The technology’s tendency to produce confident but entirely fabricated citations makes this a practical risk rather than a theoretical one.