Standing Orders and the Bench: How Judges are Policing AI in the Courtroom

The judiciary is no longer waiting for the profession to self-correct on AI. After two years of briefs filed with hallucinated citations, judges across the country, from federal district courts to state appellate panels, have started issuing specific standing orders that regulate generative AI use in court filings.

California: A Statewide Rule

The California Rules of Court established a sweeping standard in Judicial Branch Administration: Rule and Standard for Use of Generative AI (SP25-01).^{1, 2} The rule defines generative AI broadly enough to cover text, images, audio, and video outputs, and requires disclosure whenever AI has been used in the preparation of court filings.

Maryland: Chukwuemeka Mezu v. Kristen Mezu

In Chukwuemeka Mezu v. Kristen Mezu,^{3, 4} the Maryland Appellate Court addressed AI-generated fabrications directly and imposed sanctions, adding to the growing line of authority that places verification responsibility squarely on the attorney filing the document.

Louisiana: In Re: Sanctions Order of Kenney

The Louisiana Fifth Circuit's decision in In Re: Sanctions Order of Kenney^{5, 6} went further. The court sanctioned the attorney, referred her to disciplinary counsel, and accepted as reasonable opposing counsel's 7.6 hours of research to hunt down the fabricated citations. The court's response to the sanctioned attorney's argument that the citations could have been verified in "7.6 seconds" called the position an "astounding lack of awareness" of her obligations under the law.

For litigation departments, the operational answer is a mandatory verification checklist that every filing has to clear. A public sanction is a fine in the short term and a permanent line on the record in the long term.

Firms that litigate need to treat the citation-verification step as a hard gate, not an aspirational best practice. The standing orders make clear that judges have run out of patience with the alternative.