AI fake-citation cases in US courts near 1,700, up sevenfold in a year

A lawyer who has previously been sanctioned twice for misusing artificial intelligence has been rebuked again by a federal judge after including AI-fabricated quotations in a court filing against the entertainment company Roc Nation.

A growing number of lawyers are facing serious penalties for using artificial intelligence to draft court papers without checking the work. By mid-2026, courts had recorded 1,667 cases where lawyers filed fake legal citations generated by AI tools. 

AI hallucinations are getting lawyers fined

According to a global database tracked by researcher Damien Charlotin, lawyers have misused artificial intelligence in up to 1,667 court matters. This misuse is centered around adding fake AI-generated citations to official court documents. The frequency of these incidents has increased by about seven times the 230 recorded a year before. 

Recently, U.S. Magistrate Judge Jennifer Willis issued an order saying attorney Tyrone Blackburn showed a “continued pattern of behavior” involving AI after it was discovered that his filing contained quotations that did not appear in the cases he cited.

Willis wrote that Blackburn’s conduct is “an outrageous breach of his ethical and professional obligations.” He’d already been sanctioned in December for citing nonexistent cases created by AI hallucinations in a New Jersey federal court. A Pennsylvania federal judge fined him $5,000 over allegedly fabricated quotations last year.

Lawyer cites AI hallucination in Roc Nation lawsuit

In the most recent case, Blackburn filed a lawsuit on behalf of Terrance Dixon, a performer and collaborator with rapper Fat Joe. Dixon alleges the musician and his management company, Roc Nation, engaged in employment-related misconduct. Dixon claims lost wages and other harms. Roc Nation and Fat Joe have denied the allegations.

Roc Nation asked the court last month to sanction Blackburn for pursuing what it called baseless claims. It then urged the court to strike Blackburn’s opposition filing, saying it was filed too late and contained what appeared to be AI-hallucinated citations.

Blackburn attempted to double down, claiming that each of the court decisions he cited was real, though he said he paraphrased some material. Judge Willis said that Blackburn’s explanation “brazenly minimizes” his conduct.

Blackburn has said he will respond to the court’s order but maintains his denial.  

In another incident, Senior U.S. District Judge Sharion Aycock of the Northern District of Mississippi disqualified all four attorneys in a contract case after discovering that lawyers on both sides had filed hallucinated citations. The dispute itself was minor: lawyer Tom Withers had brought a single breach-of-contract claim against the city of Aberdeen over unpaid legal fees. 

Withers represented himself here only nominally and was not sanctioned. However, two of his lawyers and the two representing the city were not so lucky. Kathryn Williams admitted to using an AI research tool, and Kathleen Wilson admitted to using generative AI to draft a response, with neither verifying the results before filing. 

Wilson’s argument was that she had not known AI could fabricate cases and did not know what a hallucinated case was. 

Aycock called the defense “insufficient and incredulous.” Not only did she cancel the trial, but she also fined the attorneys between $1,000 and $3,500. Furthermore, Williams and Wilson are both barred from her district’s courts for two years. 

Where else has AI hallucinated? 

Sullivan & Cromwell, one of Wall Street’s most established firms, told a New York federal court back in April 2026 that a filing containing hallucinated errors made it into the Prince Group case. 

Andrew Dietderich, co-head of the firm’s restructuring group, apologized in a letter to Judge Martin Glenn for a series of mistakes that included misquoting the U.S. bankruptcy code and citing cases incorrectly. 

The firm said its AI policies were not followed and that a second review missed the bad citations before the filing went out. A rival firm on the case, Boies Schiller Flexner, caught them.

Sullivan & Cromwell was representing liquidators who were going after Prince Group, which was controlled by Chinese-born businessman Chen Zhi. 

U.S. prosecutors charged Chen Zhi with wire fraud and money laundering last year over what they described as forced-labor scam compounds in Cambodia. Separately, prosecutors moved to seize close to $9 billion in Bitcoin that authorities allege came from the group’s activity. Chen was arrested in Cambodia and extradited to China.

Why do AI legal tools keep making mistakes? 

Tests on two major legal research platforms, Lexis+ AI and Thomson Reuters (NYSE: TRI) AI tools, found wrong information more than 17% of the time. In some tests, error rates hit 34%. 

In February 2026, the National Center for State Courts warned that AI hallucinations are a growing threat to the justice system. 

While lawyers are not banned from using AI, they have an ethical duty to make sure everything they file is correct. 

The American Bar Association warned against failing to verify AI output back in June 2025, stating that it could violate a lawyer’s duty to provide competent representation, but these sanctions show that many lawyers have not taken that warning seriously.

The smartest crypto minds already read our newsletter. Want in? Join them.