ChatGPT Generated Fake Legal Citations in Multiple Court Cases Worldwide

Beginning in early 2023, a pattern of incidents emerged where lawyers submitted legal briefs containing fabricated case citations generated by ChatGPT to courts across multiple jurisdictions. The most prominent case involved attorneys Steven Schwartz and Peter LoDuca of Levidow, Levidow & Oberman, who submitted a brief in Mata v. Avianca containing six non-existent cases with realistic citations, judicial opinions, and legal reasoning that ChatGPT had hallucinated. Following the Avianca revelation, similar incidents surfaced internationally. In Canada, lawyers were discovered to have submitted AI-generated briefs with fabricated citations in family court proceedings. In the UK, barristers faced disciplinary action for similar conduct in both civil and criminal cases. The Park v. Kim case in New York involved another instance where ChatGPT-generated fake precedents were submitted to court, while the Kruse v. Kruse divorce proceedings in Michigan revealed extensive use of fabricated case law. Judges across jurisdictions expressed serious concerns about the integrity of the legal system. Federal Judge P. Kevin Castel in the Avianca case stated that the fabricated cases had 'bogus judicial decisions with bogus quotes and bogus internal citations.' Courts began implementing new requirements for lawyers to certify the authenticity of their citations and disclose AI assistance in brief preparation. Bar associations responded with emergency guidance and new ethical rules. The American Bar Association issued Model Rule 5.5 amendments addressing AI use, while state bars implemented continuing education requirements on AI technology. The Law Society of England and Wales published comprehensive guidance on AI use in legal practice, emphasizing the lawyer's duty to verify all AI-generated content. The incidents revealed systemic issues with AI literacy in the legal profession and highlighted the need for technological competence standards. Many lawyers admitted they were unaware that ChatGPT could generate false information that appeared authentic. The widespread nature of these incidents across multiple jurisdictions demonstrated that this was not isolated misconduct but rather a profession-wide knowledge gap about AI capabilities and limitations.