Should Clark sanction Rich, AI use still leaves legal consequences

On September twenty-fourth, the court in Clark v. Social Security, Commissioner of ordered Rich to show cause why she should not be sanctioned, including with a fine, for including factitious citations in a plaintiff’s brief. CourtListener RECAP dockets: generative AI and sanctions or hallucinations reports that Rich used generative artificial intelligence in preparing the brief and addressed that use in her response to the show-cause order. The order does not impose the sanction; it requires an explanation before the court decides what follows.

The order puts the relevant consequence in the right place: not in the novelty of the tool, but in the filing that contained citations the court called factitious. AI does not eliminate the consequences of bad legal work. A lawyer may use generative artificial intelligence and still face an order to explain the work, a possible fine, and scrutiny of the filing itself. That is the important signal for any model that presents automated drafting as finished legal work. The tool can change how a brief is produced; it does not change the court’s ability to demand that the submitted work be accurate or to impose consequences when it is not. Clark therefore exemplifies accountability surviving AI use, while leaving open the separate question of how much of the work was produced by the tool and how much was checked by the lawyer.

Other published matters point in the same direction, but they do not make the Clark outcome larger than it is. CourtListener: AI and sanctions (opinions) reports that Douglas v. Deutsche Bank National Trust Co., Published Order describes misuse of artificial intelligence as a cautionary tale about its consequences for attorneys. The same publisher reports that R. M. v. Mohr concerns fabricated material and that Cervantes v. Bianco disapproves of improper use of artificial intelligence tools that apparently contributed to a problem. Those matters show that courts can connect AI-assisted work with professional or procedural consequences. They do not establish that every AI-assisted filing is defective, that every use produces a sanction, or that an AI-native legal service has failed. Clark is one show-cause proceeding, not a finding about an entire firm or delivery model. A stronger reading would require a final sanction and repeated orders showing that the same accountability problem persists across matters. The current order does not supply either one.

The next observable fact is Clark’s final order after Rich’s response: whether the court imposes a sanction, sets a fine, or closes the matter without either. That result will distinguish a warning from an actual consequence. Another useful comparison is a filing that makes the checking obligation explicit. CourtListener RECAP dockets: generative AI and sanctions or hallucinations reports that Macias v. CCADC disclosed that OpenAI Codex assisted in preparing the document and included a certification that the filer had personally checked the accuracy of the AI-assisted portions, including all citations. A final Clark order, alongside more filings that disclose AI use and personal review, would strengthen the accountability reading. No sanction or a clear finding that the citations were accurate would narrow it. We will treat the final order, not the presence of AI by itself, as the decisive signal.