An official from the National Court Administration warns that citing AI-generated false precedents can slow court proceedings, stressing the need for effective countermeasures informed by survey results.

Nearly 92% of respondents advocate institutional countermeasures to address AI-generated fake precedents, including potential fines, verification requirements for cited material, and disciplinary actions against lawyers who repeatedly cite false information.

A large share of judges, 196 out of 243 respondents (80.7%), report encountering citations of nonexistent legal precedents, with 159 judges (65.4%) noting such cases in civil, family, and administrative matters.

The survey of 360 Supreme Court judges, conducted October 6–16, reveals that a significant majority have seen AI-generated false or non-existent precedents across various case types.

Additionally, 41 judges (16.9%) encountered citations of false or non-existent laws or regulations, and 10 judges (4.1%) saw references to false or non-existent theses or publications.

The findings highlight growing concern about AI hallucinations in legal contexts and spur consideration of formal safeguards within the judiciary and legal profession.