Mezu v. Mezu
Appellate Court of Maryland · Md. App. Ct. · Maryland bar guidance
Verified October 1, 2026
- Citation
- Mezu v. Mezu, 267 Md. App. 354, No. 361, Sept. Term 2025 (Md. App. Ct. Nov. 3, 2025)
- Decided
- November 3, 2025
Summary
In a marital dispute appeal, attorney for appellant submitted a brief containing fictitious cases, misquoted passages, and citations generated via ChatGPT, VLex, CourtListener, CaseMine, and Justia, none of which were independently verified. Judge Kathryn Grill Graeff, writing for the Appellate Court of Maryland, held that "[t]he use of AI in a legal practice is not inherently problematic" but that "[i]t is unquestionably improper for an attorney to submit a brief with fake cases generated by AI."
- AI tool:
- ChatGPT, VLex, CourtListener, CaseMine, Justia
How did the court address the AI-related conduct?
Attorney referred to the Maryland Attorney Grievance Commission. The court held attorneys cannot delegate verification responsibility and must independently review all citations before submission. First published Maryland appellate opinion directly addressing AI hallucinations in court filings; binding precedent in Maryland. The decision cited Maryland Rule 1-311(b) (attorney signature certification) and MARPC Rule 19-303.1 (meritorious claims) and is the doctrinal foundation for the proposed amendment to Rule 1-311 that would expressly codify the duty to confirm the existence and authenticity of each cited authority.
Why does Mezu v. Mezu matter for law firms using AI?
Mezu v. Mezu is the cleanest doctrinal statement from a state appellate court on the AI verification duty: AI use is fine, failure to verify is not. The opinion is binding precedent in Maryland and has already motivated a proposed amendment to Rule 1-311 that would make the verification duty explicit in the certification an attorney gives when signing a filing. The list of tools implicated, including two legal-specific platforms (VLex, CaseMine) alongside ChatGPT and two research services (CourtListener, Justia), is a reminder that hallucinations are not confined to consumer-grade AI.