Surveying 360 Judges On Fake Precedents And AI In Korea
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: Surveying 360 Judges On Fake Precedents And AI In Korea on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get smart everyday buys delivered free with Prime

  • Fast, free delivery on millions of items
  • Prime Video, Amazon Music and more included
  • Member-only deals all year
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

TL;DR

South Korea’s National Court Administration published a survey on November 2, 2026, in which judges reported seeing nonexistent or distorted legal citations in court filings. The figures describe judges’ experiences, not the share of filings containing false material; 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 many had encountered nonexistent or distorted legal citations in court filings, including material associated with AI-generated text. Published November 2, 2026, the results have prompted calls for formal safeguards; the administration plans to propose procedural law changes that would allow fines, according to the survey report.

The survey asked judges at courts of all levels to respond between October 6 and 16. Among responding judges handling civil, family and administrative cases who had encountered false citations, 196 (80.7%) identified precedents that did not exist, while 159 (65.4%) reported real precedents cited with false content. Respondents could select more than one type.

The often-cited figure of about eight in ten applies to judges in those case areas, not all 360 respondents. Kyunghyang Shinmun reported that 243 of the 360 judges (67.5%) said they had encountered false citations. The court administration’s release does not itself give that overall figure. The survey records judges’ experiences; it does not establish what proportion of court filings contain false material.

Judges identified both litigants and legal professionals as sources. 164 judges (69.5%) selected parties representing themselves, and 136 (57.6%) selected lawyers acting as representatives or defense counsel. Those answers are not mutually exclusive. The release also described examples including invented judgments formatted to resemble Supreme Court decisions, real rulings misrepresented in filings, and text from AI conversations or litigation-strategy notes left in submitted documents.

At a glance
reportWhen: Survey conducted October 6–16, 2026; re…
The developmentSouth Korea’s National Court Administration released results from a survey of 360 judges on false legal citations and proposed responses, including possible fines.
Video: Fake AI Precedents in Korean Courts: What a Survey of 360 Judges Found. Narration: AI voice clone of Thorsten Meyer; images AI-generated.

Proposed Rules for False Citations

The findings matter because false citations can impose work on opposing parties and judges, and may slow proceedings while the court checks whether a case or quotation is genuine. The administration’s release said fabricated citations create an unnecessary verification burden and can hinder the speedy progress of trials.

Judges strongly favored an institutional response: 91.7% said one was needed, and 64.8% described it as very necessary. In the measures reported, the most popular choices were a legal basis for fines (61.4%), a duty to verify citations (61.1%), and a duty to disclose AI use (60.6%). These are survey responses, not rules already in force. If adopted, any new requirements could affect how lawyers and self-represented parties prepare and check court submissions.

Products Worth Considering

FLEXISPOT 60 x 30-Inch Bamboo Electric Standing Desk with White Frame

FLEXISPOT 60 x 30-Inch Bamboo Electric Standing Desk with White Frame

Apple iMac 24-inch All-in-One Desktop Computer with M6 Chip, Blue

Apple iMac 24-inch All-in-One Desktop Computer with M6 Chip, Blue

As an affiliate, we earn on qualifying purchases.

What the Survey Actually Measured

The National Court Administration is the Supreme Court’s administrative arm. Its Judicial Support Office issued a four-page Korean-language release titled “Results of the survey on citations of AI-generated false precedents and similar material.” The release says the survey was prompted by concerns that AI systems can generate false or nonexistent precedents.

The release does not state how many judges were invited or provide the exact wording of the questions. Its percentages are based on respondents to the relevant questions, and some questions allowed multiple answers. These limits matter when comparing categories or treating the findings as a measure of the scale of the problem across all Korean courts.

A translation issue also affected some English-language coverage. Yonhap’s English version, republished by The Korea Herald, said 164 judges had cited false precedents themselves. The Korean source identifies the litigants themselves as the people judges named as a source. The distinction is important: the survey does not report judges admitting to filing false citations.

Limits of the Judges’ Responses

The survey does not establish how many filings contained false material, how often such citations occurred, or how many were produced with AI. It records whether judges said they had encountered cases, and the release does not provide the number of judges invited, the full question wording, or enough detail to assess how representative the respondents were.

It is also unclear from the published material what evidence judges used to identify each citation as false, or whether the cases involved deliberate fabrication, accidental copying, or reliance on inaccurate AI output. The administration’s proposed penalties and duties have not yet become law, and their scope and enforcement details remain unsettled.

Procedural Law Proposals Ahead

The National Court Administration says it plans to propose amendments to South Korea’s Civil and Criminal Procedure Acts that would provide a legal basis for fines. The survey release does not set out a timetable for submitting the proposals or specify the penalty amounts, who would be covered, or what process would be used to determine a violation.

Any proposal would need to proceed through the relevant lawmaking process before becoming a binding requirement. Until then, the survey reflects judges’ reported experiences and preferences, rather than a new citation-checking or AI-disclosure rule. The next developments to watch are publication of draft language and decisions on whether to adopt verification duties, AI-use disclosure, or sanctions.

Key Questions

How many judges took part in the survey?

360 judges from courts at all levels responded between October 6 and 16, 2026, according to the National Court Administration.

Does the survey say that 80% of all judges saw fake citations?

No. The roughly 80% figure applies to responding judges handling civil, family and administrative cases who reported encountering false citations. Kyunghyang Shinmun reported that 243 of all 360 respondents, or 67.5%, had encountered them.

Did the survey find that lawyers submitted false citations?

Judges could identify sources, and 136 (57.6%) selected lawyers acting as representatives or defense counsel. The survey also named self-represented parties; the categories were not mutually exclusive.

Are fines or AI-disclosure rules already in effect?

No. Judges favored measures including fines, citation verification and disclosure of AI use, but these were survey preferences. The administration says it plans to propose procedural law amendments; the survey does not say that new rules have been enacted.

Source: ThorstenMeyerAI.com

This content is for general information only and is not financial, tax or legal advice. Consult a qualified professional for decisions about your money.
You May Also Like

ShinyHunters · The New APT Model.

Analysis of ShinyHunters’ evolving operational model, highlighting its AI-enabled capabilities and impact on enterprise security since 2020.

Your Software And OpenAI Agent Training: Take A Closer Look At Ironclad

OpenAI reports training GPT-6 Astra on Ironclad contract workflows. Its 55% rubric score is not a task-completion rate, and timing estimates are simulated.

MVST Deadline: MVST Investors With Losses In Excess Of $100K Have Opportunity To Lead Microvast Holdings, Inc. Securities Fraud Lawsuit

Investors with losses over $100K in Microvast Holdings have opportunity to lead a securities fraud lawsuit, according to recent PR Newswire report.

Mayor Mamdani Announces “Click-to-Cancel” Rules

Mayor Mamdani introduces new ‘Click-to-Cancel’ regulations aimed at consumer protection, effective immediately. Details on the rules and implications inside.