Judge Sharp, sitting in New York City, learned that lesson recently when he delivered a stern warning to attorneys. He warned about the dangers of employing artificial intelligence as part of legal research. The “most distressing” of the judge’s findings was a disturbing trend where attorneys have referenced what were later found to be AI-generated falsehoods in submissions and court filings. The court therefore decided not to initiate contempt proceedings against the city in this instance. Still, Judge Sharp made clear that this ruling should not be seen as a precedent in allowing such behavior moving forward.
Judge Sharp’s worries came to head in a case involving an attorney pursuing a contingency fee on behalf of a plaintiff seeking damages from two banks. We all have to admit that the same lawyer filed 45 pages including 45 separate citations, of which 18 were determined to be nonexistent. In one particularly outrageous example, a man lost his home to eviction in London. His attorney pointed to five other successful cases that have yet to be independently corroborated.
The judge’s ruling noted that generative AI tools, such as ChatGPT, “are not capable of conducting reliable legal research.” She emphasized that these tools can generate responses that look very logical and legitimate. They can just as easily generate totally made-up falsehoods.
“Such tools can produce apparently coherent and plausible responses to prompts, but those coherent and plausible responses may turn out to be entirely incorrect,” – Judge Sharp.
For now, the court chose not to intervene any further. In addition, Judge Sharp warned that attorneys who fail in their obligations would be met with “significant ramifications.” The court has extraordinary powers at its disposal. It can publicly admonish, impose the public cost, institute contempt proceedings on its own or through referral, or refer to law enforcement.
Judge Sharp wanted to stress the importance of precision in legal work. He strongly emphasized that lawyers need to double-check everything generated by AI tools against trusted sources before using it in their practice.
“Lawyers who do not comply with their professional obligations in this respect risk severe sanction,” – Judge Sharp.
She underscored the necessity for improved guidance on the use of AI in legal research, indicating that “more needs to be done to ensure that the guidance is followed and lawyers comply with their duties to the court.”
In addition to her recommendations, Judge Sharp will now forward her findings to relevant professional bodies, such as the Bar Council and the Law Society. Her goal is to provide a clear and more effective framework for adopting AI tools into the practice of law.
