Attorneys and judges are reporting a continued rise in the submission of fabricated legal materials generated by artificial intelligence, despite several years of court-ordered sanctions. In a three-day period in September 2026, legal errors involving generative AI were reported in New Mexico, Oklahoma, and California. These incidents included the filing of fabricated police witness testimony, judicial orders containing fake case citations, and erroneous legal quotations.
The trend follows the first U.S. court sanction for AI-generated errors in 2023, issued by U.S. District Judge P. Kevin Castel in Manhattan. While lawyers are generally permitted to use AI, professional ethics rules require them to verify the accuracy of all court submissions. A researcher tracking these occurrences reported that AI "hallucinations"—a term for when AI produces plausible but false information—have appeared in at least 1,395 U.S. state and federal court cases.
Recent cases highlight the range of errors. In New Mexico, a defense lawyer in a murder appeal cited fabricated witnesses and police testimony. During a hearing in August 2026, Justice C. Shannon Bacon of the New Mexico Supreme Court expressed concern that attorneys were failing to vet the technology. In Los Angeles, a lawyer representing State Farm was fined $999.99 on September 14, 2026, for submitting seven filings with inaccurate citations and quotes. Additionally, an Oklahoma judge disclosed to an investigator that two citations in a drafted order were for non-existent cases after he used an AI program for research.
The presence of fabricated testimony and fake case law creates a risk to the reliability of legal records and the outcome of litigation. While some lawyers have received warnings or formal rebukes, others have faced monetary penalties ranging from hundreds to tens of thousands of dollars. Beyond financial costs, these incidents can lead to professional consequences, as some judges have referred attorneys to state bar authorities for potential ethics violations. The prevalence of these errors prompted U.S. Senator Chuck Grassley to call on judges to formalize measures to prevent AI misuse within judicial chambers.
What happens next depends on how courts and regulatory bodies choose to address the ongoing issue. Judge Castel noted that as the problem continues, the need for sanctions that serve as a greater deterrent grows. Individual law firms, such as Morgan & Morgan, are moving forward with large-scale investments in AI, with that firm planning a $1 billion investment over the next decade. However, the legal profession remains under pressure to balance the speed of AI adoption with the ethical requirement for human verification. Further court dates or specific deadlines for new national AI regulations were not reported.