Skip to content
Submit incident
AII-1001-50877
Under investigationA regulator, court or police has opened a formal inquiry and reached no finding yet.

California prosecutor's office used AI in legal briefs with fabricated citations

A California prosecutor's office used generative AI to draft legal briefs containing hallucinated citations and misinterpreted law, according to defense lawyers representing a defendant held without bail. The district attorney acknowledged AI was used in at least one filing and stated it is implementing policies to prevent future errors, while defense advocates petition the state supreme court to investigate the pattern and concerns about due process.

April 2025 ยท Nevada County District Attorney's Office ยท Reliability and hallucination
April 2025
AII-1001-508772 sources โ†“
Quick facts
Category
Reliability and hallucination
Country
United States
Source
The New York Times
Date
April 2025
Drafted by
Raidu Incident Detection

What happened

The Nevada County District Attorney's Office filed legal briefs containing errors typical of generative AI, including fabricated legal citations and misinterpretations of law. Kyle Kjoller's lawyers identified errors in briefs across four separate cases from District Attorney Jesse Wilson's office.

Wilson acknowledged that AI was used to draft one brief, but stated it was not the one filed in Kjoller's case. He attributed other errors to human error rather than AI use.

Reported impact

Kjoller was ordered held without bail in April despite arguing that the charges against himโ€”multiple counts of illegal gun possessionโ€”were not grave enough under California law to warrant pretrial detention. His lawyers contend the AI-generated brief's errors violated his due process rights.

A group of 22 legal and technology scholars, including Barry Scheck from the Innocence Project, warned that unchecked AI use in prosecutions could lead to wrongful convictions and represents a threat to the legitimacy of courts.

Index analysis

Kjoller's lawyers filed a petition with the California Supreme Court identifying three cases containing errors typical of generative AI, including nonexistent quotations and misinterpreted court rulings. The defense requested sanctions against prosecutors but the appeals court denied the request without explanation.

The errors included wholesale misinterpretations of law and quotations that do not appear in cited texts, hallmark mistakes of generative AI systems.

Governance controls

The District Attorney's Office stated it is learning AI's pitfalls and implementing new policies in response to the errors. Wilson said prosecutors work under heavy caseloads and time constraints, and that there was no intent to mislead the court.

Classification

Category
Reliability and hallucination
Industry
Government, security and defence
Country
United States
Harm status
Actual harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Human or fundamental rights, Public interest
Harmed
General public
Failure modes
Hallucination, Fabricated citation
AI systems
Unspecified generative AI system
Regulations
FCRA (cited)

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

The New York Times
Prosecutor Used Flawed A.I. to Keep a Man in Jail, His Lawyers Say
25 November 2025
Yahoo News UK
California prosecutors' office used AI to file inaccurate motion in criminal case
27 November 2025