Study Highlights Persistent Hallucinations in Legal AI Systems

May 23, 2024

Stanford University's Human-Centered AI Institute (HAI) conducted a study involving a pre-registered dataset of over 200 open-ended legal queries to test AI products by LexisNexis and Thomson Reuters. The researchers found that these legal models exhibit hallucinations in approximately 1 out of every 6 benchmarking queries. The incident highlights the need for responsible AI governance and trustworthy AI practices, particularly in the legal sector. For more information on Project Cerebellum's AI incident database and harm prevention strategies, go to JOIN US.

Matched TAIM controls

Suggested mapping from embedding similarity (not a formal assessment). Browse all TAIM controls

Alleged deployer
Organizations requiring legal research, Legal professionals, Law firms
Alleged developer
Thomson Reuters, LexisNexis
Alleged harmed parties
Legal system, Legal professionals, Clients of lawyers

AI governance case studies

For forensic AI governance failure analysis (TAIMScore™ case studies), browse Human Signal’s Failure Files™.

Source

Data from the AI Incident Database (AIID). Cite this incident: https://incidentdatabase.ai/cite/704

Data source

Incident data is from the AI Incident Database (AIID).

When citing the database as a whole, please use:

McGregor, S. (2021) Preventing Repeated Real World AI Failures by Cataloging Incidents: The AI Incident Database. In Proceedings of the Thirty-Third Annual Conference on Innovative Applications of Artificial Intelligence (IAAI-21). Virtual Conference.

Pre-print on arXiv · Database snapshots & citation guide

We use weekly snapshots of the AIID for stable reference. For the official suggested citation of a specific incident, use the “Cite this incident” link on each incident page.