AI Hallucinations Emerge as Primary Concern for Legal Professionals in New Research
Recent findings from a LexisNexis report, titled Scaling up: The AI dependent lawyer, indicate that AI hallucinations have become the foremost concern for legal professionals in the UK. This concern has intensified despite the widespread adoption of AI technology within the legal sector.
The report reveals that 83% of UK legal professionals express apprehension regarding reliance on inaccurate or fabricated information produced by AI. This figure marks a significant increase from 57% reported at the beginning of 2024, highlighting a growing unease about the reliability of AI generated content.
In addition to hallucinations, other notable concerns include the potential for confidential data leaks, cited by 53% of respondents, and the fear of excessive reliance on AI technology, noted by 52%. These findings emerge in a context where AI usage has reached near universal levels among legal professionals.
The survey, which included responses from 543 UK legal professionals, found that 94% now utilise AI for legal tasks. Among these users, 74% engage with AI at least once a week, and over a third, specifically 34%, use it on a daily basis.
Legal research has emerged as the most prevalent application of AI, with 69% of respondents indicating its use in this area. Other common applications include document summarisation, reported by 62% of participants, while 53% use AI for both drafting and document review. However, this increasing reliance does not appear to alleviate concerns regarding the trustworthiness of AI generated outputs.
Notably, 81% of lawyers stated they feel more comfortable using AI that is grounded in legal sources, an increase from 70% in the previous year. Furthermore, over three quarters, or 77%, of those surveyed believe that AI may lead clients or members of the public to misinterpret legal matters.
Despite these concerns, the reliance on AI continues to grow. Approximately 10% of respondents now describe themselves as dependent on AI for their work, with this figure rising to 15% among large law firms and 13% among in house lawyers.
This research follows several high profile incidents involving AI errors, including a case where Pinsent Masons cited a non existent insolvency rule due to AI generated research and another instance at US firm Sullivan & Cromwell, which issued an apology to a New York judge after fabricated citations appeared in a court filing.