AI Hallucinations Lead to Sanctions for Lawyers in US and Israel
A growing number of legal professionals are facing disciplinary actions, fines, and even disbarment due to relying on AI-generated legal precedents that do not exist. Judges are increasingly finding that complex legal arguments submitted by attorneys are based on entirely fabricated case law created by artificial intelligence tools. This trend has led to significant repercussions in both the United States and Israel.
In a recent case in Mississippi, federal judge Sharion Aycock removed both parties' attorneys from the "Withers v. City of Aberdeen" case after they submitted six completely invented legal precedents. The judge also imposed financial penalties and revoked their pro hac vice status, a temporary privilege allowing out-of-state lawyers to practice in her court. Legal experts emphasize that ignorance of AI's risks is no longer a valid defense.
Israel's Supreme Court also addressed this issue in February 2025. A lawyer representing a client in a divorce dispute was reprimanded by Justice Gila Kenfy-Stein for blindly relying on AI, citing non-existent court rulings. While the court acknowledged the novelty of the situation and refrained from financial sanctions, the justice stressed that using AI does not absolve legal professionals of their duty to verify information and that submitting fabricated content constitutes a breach of professional obligations.
Legal experts from the UK, Dr. Eden Shreid and Dr. Haim Avraham, warn that AI's uncritical use could undermine the rationalization process in law, likening it to relying on a "digital fortune teller." They also note the practical dangers, with one firm dismissing an employee for using ChatGPT to generate responses to complex legal questions, a move deemed a severe breach of professionalism.
Furthermore, the risks extend to clients. A federal court in New York ruled that information shared by a criminal defendant with the AI Claude is not protected by attorney-client privilege, as AI is not a lawyer and conversations are not confidential. This ruling implies that clients revealing sensitive case details to AI could have that information used against them in legal proceedings. Startups attempting to offer AI legal services have also faced regulatory action, with DoNotPay being fined $193,000 by the FTC for misleading claims and producing flawed documents.
