On August 12, 2026 District Judge Gloria M. Navarro of the United States District Court for the District of Nevada entered an order in Phillips v. Parlade, case number 2:25-cv-01464-GMN-NJK, dismissing the complaint with prejudice. The plaintiff had sued a state court judge, alleging that she “relied wholly on artificial intelligence to issue a judicial ruling, without any discretionary human thought.” The court did not decide whether the judge in fact used AI. It held that even taking the allegation as pleaded, the conduct was still a judicial act and therefore protected by absolute judicial immunity. The order accepts in part a Report and Recommendation from Magistrate Judge Nancy J. Koppe, and expressly declines to reach the Rooker-Feldman abstention question the magistrate had also raised.
The reasoning runs through the standard Ninth Circuit test. Under Mireles v. Waco and Meek v. County of Riverside, judicial immunity applies unless the challenged conduct was accompanied by a clear absence of all jurisdiction or was not judicial in nature, and whether an act is judicial turns on four factors: whether it is a normal judicial function, whether events occurred in chambers, whether the controversy centered on a case pending before the judge, and whether it arose from a confrontation with the judge in an official capacity. The court found that “Issuing a judicial ruling is clearly a normal judicial function,” that the dispute centered on the plaintiff’s own state court case, and that nothing alleged placed the events outside chambers. On the jurisdiction prong it noted simply that the plaintiff “provides no case law or authority” supporting a clear absence of jurisdiction.
What the order does not do matters as much as what it does. It does not rule that judges may use AI to decide cases, it does not evaluate any ethics rule or standing order on generative AI in the judiciary, and it does not create any precedent binding beyond this district. It is a three-page order on a motion posture, resolving an immunity defense rather than the underlying question. The plaintiff’s factual basis, as described in the record, was inferential rather than documentary.
The reason a business or technical leader should care is the accountability structure this exposes rather than the case itself. Courts have already been willing to sanction lawyers for AI-fabricated citations, which puts the cost of AI error on the party who filed it. This order points the other way: where a decision-maker is shielded by immunity, delegating the reasoning to a model does not pierce that shield, because immunity attaches to the role and not to the method. The same logic will surface wherever a protected or regulated decision-maker adopts AI - a licensed professional, a government adjudicator, an officer acting within statutory discretion - and the practical question for anyone deploying AI into consequential decisions is whether the resulting harm has any addressable owner at all. One district court order is a data point, not a doctrine, but it is an early marker of how existing immunity law absorbs automated decision-making without being amended.