The rapid adoption of generative artificial intelligence (AI) by attorneys has transformed legal research and drafting, but it has also introduced new professional liability risks. Since 2023, several courts have sanctioned attorneys who filed briefs containing fictitious judicial opinions or inaccurate legal citations generated by AI. Although these decisions have focused primarily on attorneys’ ethical obligations and the integrity of judicial proceedings, they also raise important questions regarding the scope of coverage under lawyers’ errors and omissions (E&O) insurance policies when clients allege malpractice arising from AI-assisted legal work.
The best-known example remains Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. June 22, 2023), in which counsel submitted a brief citing multiple nonexistent judicial opinions generated by ChatGPT. The court imposed sanctions after concluding that counsel failed to verify the authorities before filing them and continued to rely on the fictitious cases after questions regarding their authenticity were raised. Since Mata, other courts have similarly emphasized that attorneys remain responsible for verifying the accuracy of legal authorities regardless of the technology used to prepare filings.
While the sanctions decisions establish that attorneys cannot delegate their professional judgment to AI, they do not resolve the separate question of insurance coverage. To date, there appears to be no reported appellate decision addressing whether a lawyers’ E&O policy covers a malpractice claim alleging damages resulting from AI-generated hallucinations. Accordingly, existing principles governing legal malpractice insurance provide the most useful framework for evaluating future coverage disputes.
Most lawyers’ E&O policies provide coverage for claims arising from a “wrongful act” committed in the rendering of professional legal services. A wrongful act typically includes any act, error, omission, or negligent act in the performance of legal services. Viewed through that lens, an attorney’s failure to verify AI-generated citations closely resembles more traditional legal malpractice claims involving negligent legal research, failure to identify controlling authority, or other litigation errors.
The insurer’s initial analysis would ordinarily focus on whether the allegations fall within the policy’s insuring agreement and whether they potentially trigger a duty to defend. Because duty-to-defend determinations generally depend upon the allegations of the complaint rather than ultimate liability, a malpractice complaint alleging negligent use of AI could present at least a potential for coverage under many professional liability policies. The more significant coverage issues are likely to arise from policy exclusions rather than from the initial grant of coverage. Most lawyers’ E&O policies contain exclusions for dishonest, fraudulent, criminal, or malicious acts. Policies may also exclude sanctions, fines, penalties, punitive damages, or the return or disgorgement of legal fees.
Accordingly, the most important facts affecting a coverage decision likely will be an attorney’s response after discovering AI-generated hallucinations. If counsel promptly informs the court that fictitious citations were inadvertently included, withdraws or corrects the filing, and acknowledges the error, the alleged conduct remains consistent with ordinary negligence in the performance of professional services.
A different analysis may emerge if an attorney, after learning that cited authorities do not exist, continues to represent them as authentic or falsely denies the circumstances under which they were generated. Should a court ultimately find that the attorney knowingly misrepresented legal authorities or intentionally attempted to mislead the tribunal, coverage could be excluded for dishonest or fraudulent conduct.
Attorneys who incorporate AI into their practice should therefore recognize that while the technology may change how legal work is performed, it does not alter the professional duty to ensure that every filing submitted to a court is accurate, supported by existing law, and independently verified. Further, when confronted with any error, an attorney should consider the words of Paul “Bear” Bryant, “When you make a mistake, there are only three things you should ever do about it: admit it, learn from it, and don’t repeat it”.