Merton101 · Case 34 · Courts and tribunals

AI Faked Citations In More Than 2,000 Cases As Google Pitches Legal Model

Reported 8 Sep 2026 · entered the ledger 25 Sep 2026 · last checked 25 Sep 2026 · Side effect

A database tracking legal hallucinations identified over 2,000 court decisions globally where AI-generated content or fake citations were submitted to courts, with self-represented litigants accounting for the majority of incidents. Courts have spent time detecting non-existent precedent, largely responding with warnings and occasional monetary sanctions.

The institution Judiciary / Courts institution · US

The mechanism Friction removed externality scale 4 of 5, fast

Adaptation verify every submission is checked before it counts

The case as the ledger holds it

The assumption that broke: Legal citations and quoted judicial opinions submitted in court briefs represent real, existing precedent verified by the filer.
The first-order effect: AI-generated citations and hallucinations appeared in at least 2,022 court decisions worldwide between April 2023 and September 2026, forcing courts to identify fake authorities, issue warnings, and sanction filers.
Who pays: Judges, court clerks, and opposing counsel who expend time attempting to verify non-existent citations, as well as litigants harmed by ungrounded filings · group: Clerks and public staff
Scale and speed: 4 of 5 · fast

Evidence 1 page read

AOL / Daily Caller News Foundation ↗ 8 Sep 2026
The data currently consists of 2,022 cases of AI-generated content, such as fake citations, used in legal decisions across the globe from April 2023 through Sept. 5 2026, according to Charlotin's website. More than 1,300 were U.S. cases.

The second reader every claim, checked against the page

Claim	Verdict	Note
[number] Generative AI hallucinations were listed in more than 2,000 court decisions since 2023, according to a public database created by legal researcher Damien Charlotin.	supported	Page states: "Generative AI hallucinations were listed in more than 2,000 court decisions since 2023, according to a public database created by legal researcher Damien Charlotin."
[number] The data currently consists of 2,022 cases of AI-generated content, such as fake citations, used in legal decisions across the globe from April 2023 through Sept. 5 2026.	supported	Page states: "The data currently consists of 2,022 cases of AI-generated content, such as fake citations, used in legal decisions across the globe from April 2023 through Sept. 5 2026"
[number] More than 1,300 were U.S. cases.	supported	Page states: "More than 1,300 were U.S. cases."
[number] Google's AI models, specifically Gemini and Bard, were cited in a total of 10 cases, according to Charlotin's data. ChatGPT was cited over 100 times.	supported	Page states: "Google's AI models, specifically Gemini and Bard, were cited in a total of 10 cases, according to Charlotin's data. ChatGPT was cited over 100 times."
[number] Most cases resulted in a warning, with fewer than 160 cases resulting in disciplinary action, according to the data. Most fines for these incidents were between $1,000 and $5,000.	supported	Page states: "Most cases resulted in a warning, with fewer than 160 cases resulting in disciplinary action, according to the data. Most fines for these incidents were between $1,000 and $5,000."
[quote] self-represented litigants (the majority of cases in the database) tend to use [free LLMs]	supported	Page quotes Charlotin: "[A]ll things otherwise equal, free LLMs [large language models], especially those that can't search the Internet, hallucinate more as a baseline... Unfortunately, that's also the kind of models self-represented litigants (the majority of cases in the database) tend to use."