Skip to content
Submit incident
AII-1001-99917
ConfirmedA court, a regulator or the organization itself has stated the core facts, or the record cites a court or regulator source.Supporting sources are pending verification.

Generative AI causes errors and fairness concerns in US courts

Generative AI tools are increasingly used in US courts for legal research and drafting pleadings, but the systems produce numerous fabricated citations, invented case precedents, and false names. A federal judge in Los Angeles imposed $31,100 in fines and sanctions on two law firms in May 2025 for a pleading containing nine legal errors in ten pages.

May 2025 Β· law firms (Los Angeles) Β· Reliability and hallucination
May 2025
AII-1001-99917
Quick facts
Category
Reliability and hallucination
Country
United States
Date
May 2025
Drafted by
Raidu Incident Detection

What happened

Generative AI is increasingly used in US courts by judges for research, lawyers for drafting pleadings, and individuals in testimony. In May 2025, a virtual avatar of a deceased victim created by generative AI delivered victim impact testimony in Phoenix, Arizona before sentencing.

Generative AI tools produce pleadings containing fabricated case citations, invented precedents, and false names. In early May 2025, a federal judge in Los Angeles imposed fines and sanctions totalling $31,100 on two law firms over a pleading with nine legal errors in ten pages, calling the process a 'collective disaster'.

Reported impact

The widespread errors in AI-drafted pleadings undermine case accuracy and expose lawyers and litigants to sanctions. Individuals filing cases pro se using generative AI to draft pleadings experience additional errors without professional oversight.

Index analysis

Lawyers and judges noted that generative AI cannot be completely relied upon and requires verification of all research results, citing well-known examples of AI confusing cases. The National Center for State Courts observed that easier pleading drafting and submission may increase the volume of lawsuits, requiring courts to prepare for growth.

Governance controls

A Portland lawyer stated that using specialized legal interfaces like ProtΓ©gΓ© and CoCounsel for case research is preferable to general-purpose tools. Northwestern Law Professor Daniel Linna cautioned that generative AI tools must be carefully integrated into the justice system, though they could improve access and efficiency.

Classification

Category
Reliability and hallucination
Industry
Government, security and defence
Country
United States
Harm status
Actual harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Economic or property, Public interest
Harmed
Business, General public
Failure modes
Fabricated citation, Hallucination
Consequences
Fine or settlement
AI systems
ChatGPT, CoCounsel, ProtΓ©gΓ©

Sources and evidence

Sources not yet published

This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.

How sources are verifiedSuggest a source for this incident