The High Court of England and Wales has cautioned lawyers that they face severe penalties if they submit false AI-generated citations in court filings. In a recent ruling linking two separate cases, Judge Victoria Sharp warned that artificial intelligence tools like ChatGPT are not reliable sources for legal research and emphasized that attorneys bear the professional responsibility to verify all research through authoritative sources before presenting it in court.
Generative AI technologies, Judge Sharp stated, have become notorious for producing responses that appear convincing yet often consist of entirely false information. “These tools can generate responses that are coherent and plausible at first glance, but under scrutiny, prove to be entirely incorrect,” she wrote. “They frequently express confident assertions that are simply untrue.”
While Judge Sharp acknowledged that AI can be helpful in legal research, she stressed that lawyers must rigorously confirm the accuracy of any information they derive from these systems. She expressed concern about a growing number of instances, both in the UK and internationally, where false citations seemingly created by AI have found their way into official legal documentation.
In one cited incident, a lawyer representing a client seeking compensation from two banks submitted a document containing 45 citations, of which 18 were identified as completely fictitious. Moreover, many of the remaining citations did not contain the quoted text attributed to them, did not substantiate the points they were supposed to uphold, or lacked any relevant connection to the matter under consideration.
In a separate case, a lawyer who challenged an eviction case cited five court precedents that did not exist. Although this lawyer denied direct use of AI, she acknowledged that the citations might have originated from AI-generated summaries found via common online searches. The judge chose not to initiate contempt proceedings against her but clarified that such leniency would not set a precedent.
Both lawyers involved in these incidents have been referred to their respective professional regulatory bodies. Judge Sharp’s ruling outlined the wide range of sanctions available to the court against attorneys who fail in their professional responsibilities; penalties can include public reprimands, cost orders, formal contempt proceedings, and potentially referrals to law enforcement.
Judge Sharp concluded by stating firmly: “Lawyers who do not comply with their professional obligations in this regard risk severe sanction.” She has directed her judgment to professional bodies such as the Bar Council and the Law Society, urging them to reinforce their guidance regarding responsible AI usage among lawyers.
