🔍 Read the full analysis: What A Korean Judges’ Survey Reveals About AI-Linked Fake Precedents on ThorstenMeyerAI.com
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
TL;DR
South Korea’s National Court Administration published a survey of 360 judges on Nov. 2, 2026, reporting that about 80% of responding civil, family and administrative judges had encountered fake or distorted case citations. The survey records judges’ experiences, not the prevalence of false citations across all court filings; the administration says it plans to propose procedural-law changes allowing fines.
South Korea’s National Court Administration said a survey of 360 judges found that about 80% of responding judges handling civil, family and administrative cases had encountered false or nonexistent precedents in court filings. The results, published Nov. 2, 2026, have prompted calls for rules to address fabricated citations, including possible fines and duties to verify sources and disclose AI use.
The survey was conducted from Oct. 6 to 16 among judges at courts across South Korea. The court administration’s release said judges had encountered precedents that did not exist, as well as real rulings cited with false content. It described examples including AI-generated judgments formatted to resemble Supreme Court decisions and case numbers linked to reasoning that could not be found in legal searches.
The headline figure requires a specific denominator. The release’s “about 80%” finding applies to responding judges responsible for civil, family and administrative cases. The source material reports that 243 of all 360 respondents had encountered fake citations, or 67.5%; that figure is reported by the Korean daily Kyunghyang Shinmun, rather than printed in the court administration’s release. Among the judges who had encountered false material, 196 identified nonexistent precedents and 159 identified real precedents presented with inaccurate content. Respondents could select more than one type.
Judges also identified who had submitted the material. The survey records 164 judges (69.5%) naming litigants themselves and 136 (57.6%) naming lawyers acting as representatives or defense counsel. These figures concern judges’ responses about sources of false citations; they do not establish how often any individual or group submits them.
Pressure for Rules on Court Filings
The findings point to a practical risk for court proceedings: when a filing cites a ruling that does not exist, or misstates a real one, judges and opposing parties must spend time checking it. The court administration said false citations burden opposing parties with verification and can hinder the timely progress of trials. The survey does not measure delays or quantify the workload, but it documents judges’ reported exposure to the problem.
There is also a professional-responsibility dimension. The responses indicate that judges have encountered false material in filings submitted not only by litigants without legal representation, but also by lawyers. That does not establish intent or show that every error was generated by AI. It does underline why source checking matters regardless of whether a citation came from a chatbot, a human drafter or another source.
91.7% of respondents said an institutional response was needed, according to the release. The preferred measures included a legal basis for fines, a duty to verify cited precedents and a duty to disclose AI use. Such rules could clarify expectations for filings, but the survey itself does not create new obligations or penalties.
Popular right now
Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Black
KitchenAid 5.5-Qt Bowl-Lift Stand Mixer, Porcelain White (KSM55SXXX)
KitchenAid Artisan 5-Quart Tilt-Head Stand Mixer, Almond Cream
As an affiliate, we earn on qualifying purchases.
What the Survey Actually Measured
This was the National Court Administration’s first survey on fake citations, according to the source material. The court administration is the administrative arm of South Korea’s Supreme Court. Its release said AI hallucinations can produce false citations and described the potential burden on parties and the pace of proceedings.
The study asked judges about what they had encountered; it was not an audit of court documents. The release does not state how many judges were invited, and the source material says the questions’ exact wording was not provided. The results therefore cannot establish the rate of fake citations across all filings, or how that rate has changed over time.
The source material also highlights a translation discrepancy in an English-language report republished by The Korea Herald. That version said 164 judges had cited false precedents themselves. The Korean original instead says those judges identified the litigants themselves as a source. The distinction matters: the survey did not report that judges admitted submitting fabricated citations.
Limits of the Judges’ Responses
The results are based on judges’ reports of cases they had encountered, not an independent review of filings. It remains unclear how many documents contained false citations, what share of filings they represent, and how often the errors were caused by AI rather than other mistakes or deliberate fabrication. The survey also does not establish whether a false citation affected a ruling.
The release does not give the number of judges invited to participate or the full wording of the questions, limiting what can be assessed about response rates and question framing. Multiple answers were allowed for the citation types and sources, so those categories should not be treated as mutually exclusive. The figures also do not show whether judges’ experiences were concentrated in particular courts or case types.
Proposed Changes to Procedure Laws
The National Court Administration plans to propose revisions to South Korea’s Civil and Criminal Procedure Acts that would provide a legal basis for fines, according to the source material. Judges also supported citation-verification duties and disclosure of AI use. The survey results alone do not say whether those measures will be included in legislation, when a proposal will be submitted, or what penalties and duties it might specify.
Any legislative proposal would need to clarify how courts distinguish an inadvertent citation error from knowing submission of fabricated material, and how requirements would apply to parties and lawyers. Until the administration or lawmakers provide those details, the survey remains evidence of judges’ reported experiences and preferences—not a change in the rules governing filings.
Key Questions
What did South Korea’s judges report?
The National Court Administration said about 80% of responding civil, family and administrative judges had encountered false or nonexistent precedents. The survey covered 360 judges overall.
Does the survey show that 80% of court filings contain fake citations?
No. It measures judges’ reported experiences, not the share of filings containing false citations. The source material reports that 243 of 360 judges had encountered fake citations, but the survey does not provide a filing-level rate.
Did judges say they had submitted false precedents?
No. The Korean release says 164 judges identified litigants themselves as a source of false material. An English-language rendering that described those judges as naming themselves conflicts with the Korean wording.
What rules do judges want?
Respondents supported an institutional response, including a legal basis for fines, a duty to verify cited precedents and a duty to disclose AI use. These are survey preferences, not rules already in force.
What happens next?
The National Court Administration plans to propose changes to the Civil and Criminal Procedure Acts to allow fines. The timing and details of any proposal have not been specified in the source material.
Source: ThorstenMeyerAI.com
