Former DOGE staffer says ChatGPT helped feds cancel grants mentioning ‘LGBTQ+’

A former Trump administration staffer testified under oath that humanities grants referencing LGBTQ+ people were flagged for cancellation, sometimes simply because the word appeared in a project description.

Keep up with the latest in LGBTQ+ news and politics. Sign up for The Advocate's email newsletter.

The testimony from Nathan Cavanaugh, a political appointee in his late twenties who worked with billionaire Elon Musk’s Department of Government Efficiency team while detailed from the General Services Administration, offers an unusually detailed window into how the administration moved to purge diversity-related projects from the National Endowment for the Humanities.

The NEH distributes hundreds of millions of dollars in grants each year to support historical research, museums, archives, and public humanities programs across the United States.

Cavanaugh’s January deposition, recently released on YouTube, was part of a lawsuit brought by the American Council of Learned Societies, the American Historical Association, and the Modern Language Association. The groups argue that the administration unlawfully terminated NEH grants connected to scholarship on race, gender, and LGBTQ+ communities.

Related: DOGE is gone, leaving behind 300,000 fired federal workers, 600,000 USAID deaths, and more

Related: Sam Altman on AI, queerness, and being grateful for the journey he’s on (exclusive)

At the center of the dispute is a process that, according to sworn testimony, relied heavily on scanning short grant descriptions for language connected to diversity, equity, and inclusion initiatives.

Cavanaugh told attorneys that he and another DOGE team member, Justin Fox, reviewed spreadsheets listing hundreds of grants issued during President Joe Biden’s administration. Their task, he said, was to identify projects that might conflict with President Donald Trump’s executive orders targeting DEI programs.

Neither man came from the academic or humanities worlds from which they were reviewing grants.

Cavanaugh, 28, built his career in the venture-backed startup world. A former Indiana University student who left school to start companies, he co-founded the technology firm Brainbase, which manages intellectual property licensing for brands, and later launched the accounting software startup FlowFi. He joined the government through the Trump administration’s Department of Government Efficiency effort, which embedded teams of political appointees across federal agencies to identify programs for elimination or restructuring.

Fox arrived from finance. Before entering government, he worked as an associate at the investment firm Nexus Capital Management. In the administration, he served as a senior adviser at GSA and joined Cavanaugh on the DOGE “small agencies” team, which reviewed federal grants across a range of agencies.

Their work at the National Endowment for the Humanities involved reviewing hundreds of projects through the lens of executive orders aimed at eliminating federal spending tied to diversity initiatives.

Fox, in a separate online deposition, testified that certain patterns quickly emerged.

“Promoting an LGBTQ study, stipending research on gender fluidity,” he said when asked what kinds of grants tended to stand out during the review process.

He explained that grants referencing LGBTQ+ topics could be interpreted as potentially conflicting with the administration’s directives. “LGBTQ is often associated with underrepresented minority groups,” Fox testified.

Cavanaugh’s deposition offered specific examples of how that logic played out.

Related: What is DEI, what does it mean, and why are companies really getting rid of it?

Related: Federal judge blocks Trump’s anti-DEI executive orders

One grant that drew scrutiny proposed a public discussion series titled “Examining experiences of LGBTQ military service.” The program aimed to bring veterans and community members together to discuss the experiences of marginalized service members, including women, Black veterans, Native Americans, immigrants, and LGBTQ+ people.

Asked why the project had been flagged, Cavanaugh offered a blunt explanation. “Because it explicitly says LGBTQ,” he testified.

Another grant exploring the legacy of HIV and AIDS activism and prison abolition also raised concerns. Cavanaugh testified that references to queer scholarship contributed to the decision.

“We felt the latter part of the description, specifically bringing feminist and queer insights into prison abolition … gender and LGBTQ studies and so forth,” he said. “So we felt that this referenced LGBTQ and preferencing and DEI altogether.”

Fox testified that identifying those connections was a central guidepost for the review.

“There was an executive order that said to eliminate [spending] on DEI and other wasteful government spending, and that was the lens,” he said.

The depositions also revealed how the team handled the volume of grant descriptions.

Fox said he created a spreadsheet that ran grant summaries through ChatGPT to generate explanations for why a project might relate to diversity, equity, or inclusion.

The prompt asked the tool to determine whether a grant involved DEI and to produce a short explanation identifying relevant language, he said. Fox testified that the AI-generated text was meant to help summarize projects for agency leaders reviewing the recommendations.

“This was only a tool to assist in contextualizing,” he said. Still, Fox admitted that labeling a grant as DEI could lead to its cancellation. “If that's what they wanted to do, yes,” he testified when asked whether tagging a grant that way could put it on the path to termination.

The testimony also revealed how little academic expertise informed the process.

Cavanaugh admitted that he and Fox did not consult scholars or the NEH’s peer review system before identifying projects for potential cancellation. Instead, they relied largely on their own judgment while scanning grant summaries. “I think a person can have enough judgment from reading books and being well-informed outside of traditional experience to make judgment calls about obvious things like a grant that literally lists DEI in its description,” he testified.

But when attorneys asked what books informed those judgments, he conceded he had not consulted any.

“There were no books,” Cavanaugh said.

Watch part of Nathan Cavanaugh's deposition below.

- YouTube www.youtube.com