Norms Impact
Lawsuit Alleges DOGE Cancelled $349,000 HVAC Grant to Museum after ChatGPT Flagged It As DEI
The federal government allegedly replaced expert NEH grant review with a ChatGPT-driven spreadsheet to cancel awards, violating basic administrative accountability and viewpoint-neutral funding norms.
Mar 13, 2026
Sources
Summary
A federal lawsuit and discovery materials describe DOGE staff using ChatGPT outputs logged in a spreadsheet to help cancel NEH humanities grants, including an HVAC preservation award to the High Point Museum in North Carolina. Executive-branch cost-cutting was operationalized as a centralized veto over expert agency grantmaking by substituting a chatbot classification workflow for career review and documented program judgments. The practical consequence is an unaccountable, error-prone pipeline for defunding institutions based on viewpoint-coded proxies, chilling protected cultural and educational activity while destabilizing public-serving infrastructure projects.
Reality Check
When a White House-created cost-cutting unit can nullify congressionally funded grants by outsourcing eligibility judgments to a chatbot and an internal spreadsheet, our administrative state becomes a partisan weapon rather than a rule-bound institution. The precedent shifts public funding from documented criteria and expert review into opaque, contestable classifications that can be tuned to punish disfavored ideas while evading normal accountability mechanisms. This conduct reflects prosecutable corruption risk because it substitutes an unofficial decision pipeline for lawful agency process and creates a blueprint for viewpoint-based retaliation in federal spending. Normalizing that method teaches future officials that due process, records integrity, and neutral administration are optional—exactly how democratic guardrails fail quietly, then permanently.
Detail
<p>Discovery filed in federal litigation by the American Council of Learned Societies and the American Historical Association describes a Department of Government Efficiency (DOGE) review process for National Endowment for the Humanities (NEH) grants that used ChatGPT to screen grant descriptions for possible links to diversity, equity, and inclusion (DEI).</p><p>In deposition testimony cited in the filings, DOGE staffer Justin Fox said employees prompted ChatGPT to assess whether grant descriptions related to DEI and recorded the responses and explanations in a spreadsheet used to guide cancellations.</p><p>The filings describe the spreadsheet as displacing an earlier list compiled by NEH staff of grants targeted for cuts.</p><p>One entry involved an NEH award of about $349,000 to the High Point Museum in North Carolina to replace or upgrade HVAC/climate-control equipment used to preserve collections; ChatGPT classified the project as DEI-related because preservation would support broader access to diverse audiences, and the grant was later terminated.</p><p>The museum’s director stated work had begun before termination and that the museum recovered roughly 70% of the award under a termination clause.</p><p>The filings also cite other flagged grants, including a North Carolina Central University proposal for teaching materials using digital archival collections, which ChatGPT also classified as DEI-related.</p>