DOGE used ChatGPT in a way that was both dumb and illegal, judge rules

Court Decision Strikes Down AI-Powered Grant Cuts

A federal judge has determined that the Department of Government Efficiency (DOGE) broke the law by using ChatGPT to identify and cancel federal grants tied to diversity, equity, and inclusion (DEI). This controversial application of AI resulted in the elimination of more than 1,400 grants from the National Endowment for the Humanities (NEH), totaling over $100 million.

How ChatGPT Was Used

Justin Fox, a DOGE staffer, testified that he submitted grant summaries to ChatGPT with a prompt asking if the content related to DEI. Without giving the AI a specific definition of "DEI," Fox relied on ChatGPT’s brief responses to decide which grants to target. Detection codes based on protected categories such as race, religion, and sexuality were also used to label certain grants as “wasteful.”

Judge’s Findings

US District Judge Colleen McMahon ruled that DOGE’s approach violated constitutional protections, including the First and Fifth Amendments. The court found that the agency’s reliance on AI did not absolve it from responsibility, noting that government officials failed to properly evaluate ChatGPT’s conclusions. The judge ordered the restoration of the grants, stating that many targeted initiatives were directly in line with NEH’s mission.

What This Means

This case highlights risks associated with using AI in official decision-making without oversight. The ruling reinforces that AI tools cannot be used as a shortcut to bypass constitutional rights or anti-discrimination laws in government processes.

Read the original story at The Verge.