Mezu v. Mezu

Mezu v. Mezu, No. 361, September Term, 2025 (Md. Ct. Spec. App.), opinion filed October 29, 2025, reported at 2025 Md. App. LEXIS 898. State intermediate appellate court. Copy of primary document: `~/workspace/goals/ai-sanctions-case-monitoring/files/sources/D-102-mezu-v-mezu.pdf`.

Mother's counsel's nonlawyer law clerk used ChatGPT and VLex (along with CourtListener and CaseMine) to research and draft the appellate brief, producing a brief containing multiple fictitious, misquoted, and unsupported authorities. Counsel admitted at oral argument that he did not read the cited cases, saying he typically relies on summaries found in treatises or on the internet rather than reading the cases he cites.

The court held counsel's failure to verify authorities violated his duty of reasonable inquiry and candor; it declined monetary sanctions but referred counsel to the Maryland Attorney Grievance Commission. Counsel's name is not stated in the opinion, which refers only to "Mother's counsel."

(a) Read every case cited before signing the brief, or at minimum confirm each exists in a traditional database. (b) Supervise nonlawyer staff: when a clerk does the research, the attorney must independently verify the output. (c) Treat AI tools as unreliable first drafts; the opinion notes the clerk had seen no "red flags" training on hallucination risks.

---