Back to News
PolicyAI Understanding briefing

Oklahoma judge admitted citing fake ChatGPT cases, investigators say

KFOR reports that an Oklahoma judge told investigators he cited at least two nonexistent cases supplied by ChatGPT in a child-custody order. The state Attorney General’s Office said the evidence did not support criminal prosecution, while judicial discipline remains with the Oklahoma Supreme Court.

4 min readRead the linked source
Source-provided image accompanying Oklahoma judge admitted citing fake ChatGPT cases, investigators say
Source referenceSource recorded
Publisher
wfla.com
Source link
wfla.comhttps://www.wfla.com/news/national/oklahoma-judge-admitted-to-citing-fake-chatgpt-cases-in-court-order-investigators-say/amp/
Source type
Linked source — primary-source status has not been established.
ContextUnderstand this in 60 seconds

Start here

Key terms

Generative AI
AI systems that produce new content such as text, images, audio, video, or code.
Citations
References to source passages or documents included in a model's response to support its claims.
Test yourselfAI Ethics Quiz

What happened

According to Nexstar’s KFOR, Stephens County District Judge Lawrence Wheeler admitted to investigators that he cited at least two nonexistent Oklahoma cases in a court order after finding them through ChatGPT. The order concerned a request for psychological testing in a child-custody case. KFOR reports that Wheeler said he wrote the order himself and used AI only for research.

Nexstar’s KFOR reports that the incident began in November during a Stephens County child-custody case. Judge Lawrence Wheeler denied an attorney’s request for the other parent to undergo a psychological test and cited several supposed Oklahoma precedents in the order. KFOR reports that at least two of those cases did not exist.

The Oklahoma Council on Judicial Complaints asked the Oklahoma State Bureau of Investigation to investigate. In an August letter to the Attorney General’s Office, Stephens County District Attorney Jason Hicks reportedly wrote that Wheeler told OSBI agents that ChatGPT had supplied at least two nonexistent cases. Wheeler said he ultimately wrote the order himself and used AI for research.

KFOR reports that the Attorney General’s Office reviewed the OSBI findings but concluded that the evidence did not support criminal prosecution. The office also said that the Oklahoma Supreme Court has ultimate authority over judicial discipline. KFOR contacted Wheeler’s attorney, who declined to comment.

This account relies on KFOR’s reporting and statements attributed to the district attorney, the Attorney General’s Office, and attorney Tim Gilpin. The source does not independently publish the court order, investigative file, or an interview transcript, so the underlying admission and the full scope of the fabricated citations are not independently confirmed here.

Source details: wfla.com

Why it matters

The report illustrates a consequential failure mode when generative AI is used in legal research: fabricated authorities can enter decisions that affect people’s rights and family relationships. KFOR’s account also shows that criminal liability and judicial discipline are separate questions. The Attorney General’s Office said the evidence did not support prosecution, while the Oklahoma Supreme Court retains ultimate authority over disciplining judges. The source does not establish whether the order was withdrawn, corrected, or reviewed on appeal.

Legal citations are verifiable factual authorities, not merely suggestions. If nonexistent cases appear in a judicial order, the problem can affect the reasoning available to the parties and complicate later review, even when the judge—not the software—signs the order.

The report separates two forms of accountability. The Attorney General’s Office found insufficient evidence for criminal prosecution, but that decision does not resolve whether judicial standards were violated. Any disciplinary response would fall under the Oklahoma Supreme Court’s authority, according to KFOR’s account.

The practical lesson reported by KFOR is that AI-assisted legal research requires source-by-source verification. The article does not establish whether the cited inaccuracies changed the custody ruling or caused measurable harm. It also does not report whether the court has introduced new safeguards.

What to watch next

The key unanswered issues are whether the Oklahoma Supreme Court will pursue discipline, whether the court order will be corrected or vacated, and whether Oklahoma courts will adopt clearer rules for verifying AI-assisted legal research. KFOR’s report does not provide the order, the OSBI investigative findings, or an independent review of the cited cases, so those details remain unconfirmed.

Watch for any action by the Oklahoma Supreme Court, including a disciplinary proceeding or a public explanation of how judges may use generative AI. The source does not report that such action has occurred.

Watch for corrections to, or further litigation over, the child-custody order. KFOR does not say whether the order remains in force, was amended, or was challenged because of the citations.

More documentation would clarify the case: the order itself, the OSBI findings, the identities and status of the nonexistent cases, and any response from Judge Wheeler. None is independently available in the supplied source.

Related guides & quizzes

AI EthicsChatGPT & LLMsPrompt EngineeringTest what you know — try a free AI quizLook up an AI term in our glossary
Found this useful?