🔍 Read the full analysis: AI, Legal Errors, And Fake Precedents: Korean Judges Share Their Views on ThorstenMeyerAI.com
Get the little things that make your day delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
South Korea’s National Court Administration published survey results on Nov. 2 showing that about 80% of responding civil, family and administrative judges had encountered false or nonexistent legal citations. The survey records judges’ reports, not the share of court filings containing errors; judges called for institutional measures including fines, citation checks and disclosure of AI use.
South Korea’s National Court Administration published survey results on Nov. 2 showing that about 80% of responding civil, family and administrative judges had encountered cases involving nonexistent or distorted legal citations. The findings have prompted calls among judges for rules addressing citation checks and artificial-intelligence use, but they measure judges’ reported experience—not the proportion of filings containing false material.
The court administration, an administrative arm of the Supreme Court, surveyed judges serving at courts of all levels from Oct. 6 to 16. A total of 360 judges responded. The administration’s release said false citations generated or assisted by AI can impose extra verification work on opposing parties and impede the timely progress of trials.
Among the judges counted in the release’s civil, family and administrative finding, 196 respondents (80.7%) reported encountering precedents that did not exist. A further 159 (65.4%) said they had seen real precedents cited with false content. The survey allowed multiple answers, so categories can overlap. It also recorded reports of false or nonexistent statutes and academic works.
The headline figure needs its denominator. The court release’s roughly 80% applies to responding judges handling civil, family and administrative cases. The Korean daily Kyunghyang Shinmun reported that 243 of all 360 respondents (67.5%) had encountered fake citations. The release itself did not print that all-respondent percentage; the figures are consistent with its category totals. Neither number is a count of the filings affected.
Pressure for Citation Safeguards
The findings matter because fabricated legal authority can mislead parties and consume court time, even when it appears in only one filing. Judges reported examples that included purported Supreme Court rulings formatted to look authentic, case numbers attached to reasoning that could not be found, and real decisions represented as saying something different from their actual holdings.
The survey also points beyond errors made by people representing themselves. In response to a question about who submitted false material, 164 judges (69.5%) named the parties themselves, while 136 (57.6%) named licensed lawyers. The results therefore put responsibility for checking citations in focus across the legal profession, not only among people using AI without legal representation.
Judges broadly favored an institutional response: 91.7% said one was needed, including 64.8% who described it as very necessary. The most frequently selected proposals were a legal basis for fines, a duty to verify citations and a duty to disclose AI use. Those preferences are survey responses, not laws already in force or proof that any particular sanction will be adopted.
Products Worth Considering
Hyperice Hypervolt Go 3 Portable Percussion Massage Gun with Carry Case
Hyperice Hypervolt 3 Pro Percussion Massage Gun with Heated Head and Carry Case
Fitbit Sense 2 Smartwatch – Blue Mist / Pale Gold
As an affiliate, we earn on qualifying purchases.
How the Survey Was Framed
The National Court Administration titled its four-page release “Results of the survey on citations of AI-generated false precedents and similar material.” It said the inquiry was prompted by concerns that AI hallucinations can produce citations to nonexistent rulings. The survey asked judges about what they had encountered; it did not independently audit court records or test whether AI caused each reported error.
The release does not state how many judges were invited to take part or provide the precise wording of every question. Percentages are based on respondents to the relevant questions, and some items allowed multiple answers. Those limits make the findings useful as a measure of judicial experience and views, but not as a nationwide rate of false citations in filings.
The administration has said it plans to propose amendments to South Korea’s Civil and Criminal Procedure Acts to establish a basis for fines. The source article also described sanctions or referrals involving lawyers in New York and London, but those examples do not establish that South Korea has adopted equivalent rules.
Limits of the Evidence
The survey does not establish how often false citations appear in all court filings, how many cases were affected, or whether the cited errors were produced by AI rather than another source. It records judges’ observations, and the administration has not published the number invited to respond or the full question wording, limiting assessment of response rates and survey design.
Some reported percentages refer to different groups of respondents, while several questions allowed multiple selections. The English-language version of the findings also requires care: according to the source material, an English news report rendered 164 judges as having cited false precedents themselves, but the Korean original says those judges identified the litigants themselves as the source. The translated claim should not be read as judges admitting misconduct.
It remains unclear what form any proposed legal amendments will take, what conduct could draw a fine, and whether rules would require disclosure of all AI use or only particular uses in preparing court documents.
Proposed Rules Still Ahead
The National Court Administration plans to propose changes to the Civil and Criminal Procedure Acts that would provide a legal basis for fines, according to the source material. The survey also records support for requiring parties or their representatives to verify cited precedents and disclose AI use, but it does not say that those proposals have been drafted or approved.
Any change would need to move through South Korea’s legislative process. Until the administration releases a proposal or lawmakers take action, the scope, enforcement and timing of new requirements remain unsettled. Further details about the survey’s methodology would also help clarify how broadly its findings apply to court filings.
Key Questions
What did South Korea’s judges report?
In the survey results published Nov. 2, about 80% of responding civil, family and administrative judges said they had encountered cases citing nonexistent or false precedents. The figure describes judges’ reported experience, not the percentage of filings containing errors.
How many judges took part?
360 judges from courts at all levels responded to the survey, which was conducted from Oct. 6 to 16, 2026.
Did judges say lawyers were responsible for false citations?
The survey asked judges who had submitted false material. 136 respondents (57.6%) identified a lawyer acting as a representative or defense counsel, and 164 (69.5%) identified the parties themselves. The question allowed multiple answers.
Are new fines or AI disclosure rules already in effect?
No. The administration plans to propose amendments that could provide a basis for fines. The survey also found support for citation-verification and AI-disclosure duties, but no such changes are established by the survey itself.
Does the survey prove AI caused the reported errors?
No. The administration cited AI-generated hallucinations as a concern, but the survey reports what judges encountered and does not determine the source of every false or distorted citation.
Source: ThorstenMeyerAI.com
