Federal judge Colleen McMahon issued a 143‑page opinion condemning appointees Justin Fox and Nate Cavanaugh for using ChatGPT to decide whether National Endowment for the Humanities (NEH) grants violated a DEI standard. Their method asked the model to answer “Yes” or “No” in under 120 characters, then canceling the awards based solely on that output and raising constitutional free‑speech concerns.

Because the prompt ignored project purpose, methodology, or scholarly merit, ChatGPT routinely flagged innocuous research—such as multispectral imaging of ancient Jewish texts or a study documenting Uyghur repression in China—as DEI. The judge called the approach “arbitrary and capricious,” noting that neither Fox nor Cavanaugh defined DEI for the model and that the NEH’s statutory review process was completely bypassed.

McMahon concluded that the agency lacked authority to revoke grants without individualized expert evaluation, and that using a generative‑AI tool cannot satisfy legal due process. The ruling reinforces that AI‑generated classifications are insufficient for statutory decision‑making and restores the traditional peer‑review safeguards that protect federally funded humanities research. Future courts will likely scrutinize similar AI‑driven administrative actions to ensure compliance with statutory mandates.