LLMMedium
Study Highlights Persistent Hallucinations in Legal AI Systems
- Occurred
- May 23, 2024
- Published
- May 31, 2024
- Source
- AI Incident Database
Summary
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.
Developers
- Thomson Reuters
- LexisNexis
Deployers
- Legal professionals
- Law firms
- Organizations requiring legal research
Harmed parties
- Legal professionals
- Clients of lawyers
- Legal system
Classification
Category confidence: 95%
Severity confidence: 90%
Rationale: Study documents hallucinations in legal AI models, a core LLM failure mode, without evidence of physical harm or legal action.
Classifier: Qwen/Qwen3.5-35B-A3B-FP8