🔍 Read the full analysis: How Widespread Are Fake AI Precedents In Korean Courts? A Survey Of 360 Judges on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
South Korea’s National Court Administration published a survey of 360 judges reporting that about 80% of responding civil, family and administrative judges had encountered fake or distorted citations. The figures describe judges’ reported experience, not the share of court filings containing false citations; the administration says it plans to propose legal 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 filings citing false or nonexistent precedents. The results, published November 2, describe judges’ reported experiences—not the proportion of filings containing fabricated citations—and the administration says it plans to propose procedural-law changes that would allow fines.
Judges at courts across the country answered the survey between October 6 and 16, 2026. In its four-page release, the National Court Administration said AI-generated hallucinations can produce false precedent citations, placing an unnecessary verification burden on opposing parties and potentially slowing trials. The release did not state how many judges were invited or provide the exact wording of the questions.
Among judges who reported encountering false citations, 196 respondents (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. Other reported material included false or nonexistent statutes, cited by 41 judges (16.9%), and false or nonexistent papers or literature, cited by 10 (4.1%).
The denominator matters. The reported 80.7% applies to judges who had experienced the problem, not all 360 participants. South Korean newspaper Kyunghyang Shinmun reported that 243 of all 360 respondents (67.5%) had encountered fake citations. The court release’s central “about 80%” finding, by contrast, concerns responding judges handling civil, family and administrative cases. Neither figure measures how often false citations appear in court documents overall.
Proposed Penalties for False Citations
The survey suggests the issue is not limited to people representing themselves. Asked who submitted false material, 164 judges (69.5%) selected the parties themselves, while 136 (57.6%) selected a lawyer acting as litigation representative or defense counsel. These results describe judges’ reports and do not establish how often either group was responsible across all cases.
91.7% of respondents called for an institutional response. The leading options were a legal basis for fines (61.4%), a duty to verify cited precedents (61.1%) and a duty to disclose AI use (60.6%). If lawmakers adopt such rules, they could make lawyers and litigants more accountable for checking authorities before submitting them. The survey itself, however, does not establish that a particular citation was produced by AI or that AI use was the cause in every case.
False authorities can complicate a proceeding even when they are caught: judges and opposing parties must spend time checking whether a cited case exists and whether it says what a filing claims. The administration said that burden can interfere with the timely progress of trials. The reported figures point to a concern inside the judiciary, while leaving the scale of the problem across filings unmeasured.
Products Worth Considering
DJI Mic 3 Wireless Microphone Kit (2 Transmitters, 1 Receiver, Charging Case)
Apple Watch Series 8 GPS, 41mm, Midnight Aluminum Case with Midnight Sport Band (S/M)
Apple Watch SE (2nd Gen) GPS 40mm with Starlight Aluminum Case and Sport Band (S/M)
As an affiliate, we earn on qualifying purchases.
What Judges Reported Seeing
The National Court Administration, the administrative arm of South Korea’s Supreme Court, titled its release “Results of the survey on citations of AI-generated false precedents and similar material.” It described examples including fabricated judgments formatted to resemble Supreme Court decisions, case numbers attached to reasoning that could not be found in searches, and real rulings whose holdings were altered or presented out of context.
The court also listed filings that retained text from conversations with an AI assistant, including wording about litigation strategy. Those examples illustrate different kinds of problems: a citation may be invented, a real authority may be misrepresented, or chatbot text may be copied into a filing without adequate review. The survey does not quantify how many incidents fell into each category or independently establish their cause.
A translation issue in an English-language report also shows why source wording matters. Yonhap’s English version, republished by The Korea Herald, said 164 judges “themselves have cited false precedents.” The Korean original instead says those judges identified the parties themselves as a source of false material. The survey does not report that the judges admitted submitting fabricated citations.
Limits of the Survey Evidence
The survey measures what judges said they had encountered, not the number or proportion of filings containing fabricated or distorted citations. The administration did not publish how many judges were invited, the response rate, or the full question wording. That makes it difficult to assess how representative the 360 responses are.
The figures also do not show how many reported citations were generated by AI, as opposed to fabricated or misstated for another reason. The release lists AI-related examples but does not provide case-level evidence, a time series or a breakdown by court, filing type or source of the material. The cited percentages also have question-specific denominators, and some questions allowed multiple answers.
It remains unclear what precise offenses or procedures any proposed fines would cover, what safeguards would apply, and whether disclosure of AI use would be mandatory. No legislation has been enacted on the basis of this survey in the material available here.
Court Administration Plans Legal Proposals
The National Court Administration said 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 give a timetable for submitting draft amendments or describe their proposed wording.
The administration’s stated next step is a proposal, not an adopted rule. Any change would require further legislative action, and the survey does not settle how courts would verify suspected false citations or apply penalties. Further details about the draft, its scope and the process for reviewing AI-related filings have not been provided.
Key Questions
What does the survey’s “about 80%” figure mean?
It refers to responding judges handling civil, family and administrative cases who said they had encountered false or nonexistent precedents. It does not mean that 80% of all court filings contained fake citations.
How many judges took part?
360 judges from courts at all levels answered between October 6 and 16, 2026. The National Court Administration did not state how many judges were invited.
Did the survey show that lawyers submitted false citations?
Judges could identify who had submitted false material. 136 respondents (57.6%) selected a lawyer acting as litigation representative or defense counsel; 164 (69.5%) selected the parties themselves. These are survey responses, not a measure of responsibility across all cases.
What action is South Korea’s court administration proposing?
It says it plans to propose amendments to the Civil and Criminal Procedure Acts to allow fines. The release does not give a submission date or the proposed rules’ exact terms.
Source: ThorstenMeyerAI.com
