The Use of Artificial Intelligence in Courts: A Warning
Civil litigation often hinges on costs rather than merits. In England and Wales, the losing party generally pays the winning party's legal costs, usually around 70% on a standard basis. Deviations from this norm signal significant issues. Judges may reduce costs or order indemnity costs if parties behave improperly. Even rarer are wasted costs orders, where lawyers are personally liable for costs due to misconduct.
The recent case of R (Ayinde) v London Borough of Haringey stands out as an example of such misconduct. This case began as a typical judicial review by a homeless claimant against a local authority. The defendant council performed poorly, ultimately being disbarred from defending the claim after breaching court orders. Thankfully, the claimant secured accommodation and the legal claim had merit.
What made this case exceptional was the conduct of the claimant’s lawyers. They submitted legal arguments supported by five fabricated cases, including one purported Court of Appeal decision. These cases looked authentic with proper citations but did not exist in any law reports. The judge noted that these fake cases served the same purpose as genuine ones would have.
Surprisingly, there was no apparent need to invent these cases. The legal points were straightforward and could have been supported by existing authorities. As the judge observed, the problem wasn’t the legal argument but the fact that the cited case “Ibrahim” was entirely fictional.
The reasons behind this fabrication remain unclear. The council speculated that the lawyers might have relied on artificial intelligence, such as ChatGPT or another large language model (LLM), for their research. The judge did not need to determine this but found it the most plausible explanation. If true, this points to the dangers of uncritical reliance on AI-generated legal content.
Legal research is fundamental to lawyering—identifying rules, applying precedents, and distinguishing cases. It cannot be delegated or rushed. While AI tools may seem like convenient search engines, they currently lack the reliability required for serious legal work. Lawyers must verify any AI-generated information thoroughly.
However, proper legal research is often time-consuming and expensive. Comprehensive legal databases charge high fees, and free resources may be incomplete. Law firms and solo practitioners face financial and time constraints, while clients expect lawyers to know the law without additional costs. This environment creates a temptation to use AI for research despite the risks.
The ruling in R (Ayinde) v London Borough of Haringey sends a clear message: using AI-generated research without proper verification can lead not only to wasted effort but also to personal cost sanctions against lawyers. This serves as a strong warning against relying on AI for legal research in its current form.
Lawyers should approach AI tools cautiously and maintain rigorous standards in legal research. The integrity of legal practice depends on it.
For those interested in understanding how to responsibly integrate AI tools into legal workflows, exploring vetted training resources can be valuable. Check out Complete AI Training’s ChatGPT courses for guidance on effective and safe AI use.
Your membership also unlocks:
