AI incident ·
Study Highlights Persistent Hallucinations in Legal AI Systems
In brief
An AI system built by Thomson Reuters and Lexisnexis and deployed by Legal Professionals, Law Firms and 1 other allegedly harmed Legal Professionals, Clients Of Lawyers and 1 other.
- Risk domain
- Misinformation
- Occurred
- Coverage
- 2 reports
What happened
Stanford University’s Human-Centered AI Institute (HAI) conducted a study in which they designed a "pre-registered dataset of over 200 open-ended legal queries" to test AI products by LexisNexis (creator of Lexis+ AI) and Thomson Reuters (creator of Westlaw AI-Assisted Research and Ask Practical Law AI). The researchers found that these legal models hallucinate in 1 out of 6 (or more) benchmarking queries.
Laws that address this harm
Policy angle: Classified under Misinformation (False or misleading information) in the MIT AI Risk Repository taxonomy; 5 recorded instruments address this use case.
- India DPDP Act
- Law on Artificial Intelligence (2025)
- Law No. 132/2025 on artificial intelligence
- EU AI Act
- Texas Responsible AI Governance Act (TRAIGA)
Matched from the record's risk domain and country to the instruments recorded here. A reviewer can correct the match in the repository (data/external/incident_overrides.yaml).
News reports (2)
Titles link to the original publisher; report text is not reproduced here.
Who was involved
- Alleged developer
- Thomson Reuters, Lexisnexis
- Alleged harmed party
- Legal Professionals, Clients Of Lawyers, Legal System
Classification (MIT AI Risk Repository taxonomy)
- Risk domain
- Misinformation
- Risk subdomain
- 3.1 False or misleading information
- Causal entity
- AI
- Intent
- Unintentional
- Timing
- Post-deployment
- Harm level
- —
- Sectors
- —
- Countries
- —
Risk entries describing this failure mode
Entries from the MIT AI Risk Repository coded to subdomain 3.1.
- Pursuing Consistent Context
"LLMs have been demonstrated to pursue consistent context [129]–[132], which may lead to erroneous generation when the prefixes contain false information. Typical examples include sycophancy [129], [130], false demonstra...
- Defective Decoding Process
In general, LLMs employ the Transformer architecture [32] and generate content in an autoregressive manner, where the prediction of the next token is conditioned on the previously generated token sequence. Such a scheme...
- Noisy Training Data
"Another important source of hallucinations is the noise in training data, which introduces errors in the knowledge stored in model parameters [111]–[113]. Generally, the training data inherently harbors misinformation....
- Hallucinations
"LLMs generate nonsensical, untruthful, and factual incorrect content"
- Faithfulness Errors
"The LLM-generated content could contain inaccurate information" which is is not true to the source material or input used
- Factuality Errors
"The LLM-generated content could contain inaccurate information" which is factually incorrect
- Untruthful Content
"The LLM-generated content could contain inaccurate information"
- Knowledge Gaps
"Since the training corpora of LLMs can not contain all possible world knowledge [114]–[119], and it is challenging for LLMs to grasp the long-tail knowledge within their training data [120], [121], LLMs inherently posse...
Incidents in the same risk subdomain
- Nonfiction Book 'The Future of Truth' Reportedly Included AI-Generated and Misattributed Quotations
- Claude Console Reportedly Generated Phantom Legal Quotations in Trump Layoffs Court Filing
- Purportedly AI-Enhanced Images of Iranian Women Protesters Were Reportedly Spread With Unverified Execution Claims
- South Africa Draft National AI Policy Reportedly Included Fictitious References Believed to Be AI Hallucinations
- Purportedly AI-Generated Image Reportedly Misled Daejeon Authorities Searching for Escaped Wolf Neukgu
- Gemini and Grok Reportedly Misidentified Authentic Minab School-Strike Graveyard Photo as Unrelated Disaster Imagery
Source record: incident #704 on the AI Incident Database · all 2 reports