In brief
The Supreme Court surveyed 360 judges from October 6‑16 and found that about 80% have seen AI‑generated fake precedents cited in civil, family and administrative cases. Ninety‑two percent of respondents urged institutional measures, including fines and verification requirements, to curb the problem. The findings highlight how AI hallucination can delay trials and erode public trust in the judiciary.
On Nov. 2, the Korean Supreme Court released the results of a survey involving 360 judges that shines a light on a growing challenge within South Korea’s courts: artificial intelligence creating and feeding judges with false legal precedents. The survey, conducted from Oct. 6 to 16, found that eight in ten judges had encountered cases where AI‑generated documents were cited as authentic precedents. This startling figure points to a rapid rise in what experts call “AI hallucination,” the phenomenon where AI models present invented facts as factual.
The study covered judges who preside over civil, family and administrative matters, the fields that see the greatest volume of cases in South Korea. Among those judges, 196, or 80.7 percent, said they had come across nonexistent precedents that were cited in court filings. A slightly smaller group—159 judges, or 65.4 percent—reported cases where true precedents were misquoted or fabricated. While the numbers differ, the overall picture is that most judges are now encountering legal arguments that rest on shaky or false foundations.
Beyond cases of nonexistent rulings, the survey also identified instances in which courts faced citations of forged legal texts. Forty‑one judges, or 16.9 percent of those surveyed, reported encountering false or nonexistent statutes and regulations. A minority—10 judges, or 4.1 percent—reported seeing unsubstantiated theses and scholarly publications cited as legal authority. These numbers indicate that AI is not limited to generating sole facts but is also capable of fabricating entire sections of law that can mislead legal practitioners.
But the survey also uncovered how the problem spreads from court staff to litigants. One hundred and sixty‑four judges admitted that they themselves had cited false precedents in court, while 136 judges, or 57.6 percent, reported that lawyers representing parties had introduced fabricated or incorrect citations. This finding suggests that the problem is not confined to a few rogue judges but is a widespread issue that could influence the outcome of many cases.
More than 90 percent of respondents—92 percent, to be exact—called for institutional measures to address the proliferation of AI‑generated fake precedents. They suggested a number of options, such as creating legal frameworks that allow the imposition of fines on parties that disseminate false citations, establishing verification requirements that would compel lawyers to confirm the authenticity of cited material, and building disciplinary mechanisms that would sanction repeated offenses by attorneys who knowingly cite fabricated information.
According to an official from the National Court Administration, citations of AI‑generated false or nonexistent precedents can delay proceedings and erode public confidence in the judicial process. The official warned that the resulting uncertainty can compromise the due‑process guarantees that underlie South Korea’s democracy. The risk is not purely procedural—it also threatens the broader principle that every individual and business should be entitled to a fair trial based on genuine legal authority.
While the survey’s findings underscore a technological problem, they also reveal a deep civic stakes dilemma: the integrity of the courts, the fairness of elections that rely on a trustworthy legal system, and the precise functioning of public institutions that depend on clear legal precedents. If judges or lawyers are misled by fabricated documents, the decisions they render might be based on falsehoods, potentially causing injustice to the litigants and causing a ripple effect that undermines the rule of law.
In response to the survey, Supreme Court officials have pledged to investigate the matter further and to draft “institutional measures” that would formally address the issue. As AI tools become more sophisticated and widely used in the legal field, the judicial system will need clearer rules and more robust verification processes to ensure that the courts remain a pillar of public trust and accountability in South Korea’s democracy.
