A court requires a lawyer to explain AI-assisted citations

On September twenty-four, in Clark v. Social Security, Commissioner of, a court ordered Rich to show cause why she should not be sanctioned, including with a fine, for including factitious citations in a plaintiff’s brief. The CourtListener RECAP filing says Rich used generative AI in preparing the brief and responded to the order. The immediate event is not a warning about a hypothetical failure: it is a court process requiring an explanation for citations that appeared in an actual filing.

AI does not eliminate the consequences of bad legal work. It changes neither the court’s demand for a supportable citation nor the lawyer’s exposure when a brief contains one that is factitious. Clark makes that point in the narrowest useful way: AI assistance appears in the preparation process, while the court’s response attaches to the filed work. The filing does not say that generative AI itself is sanctionable. It says that a person was ordered to explain a brief containing factitious citations after AI was used in preparing it. That distinction is the accountability mechanism an AI-native legal service must preserve.

Clark proves that one AI-assisted brief can produce a show-cause order and possible financial sanctions. It does not prove that every AI-assisted filing is unreliable, that every citation error results from AI, or that a court has already imposed a sanction in this case. The stronger reading would require a final order connecting the citations, the use of AI, and an actual penalty. That has not happened in the material describing the September twenty-four filing. A separate report from Legaltech News illustrates why the boundary matters: a newly filed lawsuit in the Superior Court of New Jersey, Gloucester County concerns an AI-powered meeting tool that allegedly captured and distributed information in ways users never intended. That report concerns information handling, not a defective legal brief, so it broadens the category of AI-related exposure without proving the Clark allegation.

The next decisive fact is a final order in Clark v. Social Security, Commissioner of. An order imposing a fine or another sanction for the factitious citations would confirm that AI-assisted legal work can carry a direct court consequence even after the work is filed. An order declining sanctions, or finding that the citations did not justify a penalty, would narrow the lesson without erasing the obligation to check the brief. The same test applies to AI-native legal providers: a finished filing is not validated by the system that helped produce it. The observable question is whether a responsible human process catches unsupported citations before they reach a court.