"Fabricated Judgments," "Deception of the Court," "Serious Illegality"... Judges Express Anger over 'AI Fiction' in Written Rulings

"The plaintiff appears to have intentionally cited false precedents in an active attempt to deceive the court, and the illegality of this conduct seems to be extremely serious. (...) Given that the plaintiff's attempts to mislead the court are ongoing, the motion to reopen the argument cannot be accepted under any circumstances."

This statement was included in the written judgment rendered by the Seoul Eastern District Court on June 12, 2026, in a civil case. The court included such remarks in the footnotes of its rulings after the issue of cases citing fabricated precedents created by AI in litigation documents became a serious concern.

The Law Times.

There have also been instances where the court explicitly pointed out that the precedents were generated by AI, moving beyond simply using phrases such as "unverifiable precedent" or "non-existent precedent." For example, in a monetary dispute on May 28, the Suwon District Court wrote in the body and footnotes of its judgment, "All of the rulings presented by the defendant appear to be fabricated judgments produced by the use of AI."

Although these particular cases were 'pro se litigation,' in which parties represented themselves without legal counsel, the issue of submitting fabricated precedents has also been identified in cases where lawyers act on behalf of clients.

On June 11, the Patent Court clearly stated in a ruling on an invalidation action brought by Company A against another company that "A considerable number of the precedents cited by the plaintiff do not exist, and submitting such unverified judgments to the court is inappropriate." In this case, the plaintiff was represented by a lawyer with 27 years of experience.

This attorney submitted a motion to reopen the argument, attaching eight fabricated precedents created by AI without verifying them. The panel also noted in its ruling that "The plaintiff admits to submitting some judgments generated via AI without verification."

As the problem of AI-generated fake precedents escalates, the Judicial Research and Training Institute held a training session for judges on September 4 under the theme "Responses to Citations of Fabricated Laws and Precedents Using AI." During the session, a case was shared in which 39 out of 40 precedents cited in the written filings by a party were determined to be falsified.

There are further reasons why courts are now explicitly stating this in their rulings. Sometimes, parties to lawsuits believe fabricated precedents created by AI to be factual and consequently refuse to accept the judge's decision. A judge in the Seoul metropolitan area explained, "Declaring in the judgment that a precedent is fabricated serves to make clear that content generated by AI may differ from actual facts."

Meanwhile, judges have analyzed that the use of AI to easily prepare litigation documents is also impacting the number of cases being filed with the courts—particularly in simple civil small claims cases.

From January to July of this year, the number of first-instance small claims cases filed amounted to 334,674,

which is an increase of approximately 12.7% compared to the same period in 2025.

A sitting judge commented, "In small claims cases, the amount in dispute is low and most cases are concluded in a single hearing, so parties have found it easier than before to file lawsuits without an attorney by utilizing AI." According to the National Court Administration, the rate of attorney representation in civil main actions in 2025 was about 87% for civil panel cases and about 56% for single-judge cases, while it was only about 22% for small claims.

Mina Han, The Law Times Reporter

※This article is based on content supplied by Law Times.

This content was produced with the assistance of AI translation services.