Sign up for Executive Dysfunction, a newsletter that highlights one under-the-radar story each week about how Trump is changing the law—or how the law is pushing back. You’ll also receive updates on the latest from Slate’s Jurisprudence team.
One year ago, the Trump administration canceled more than 1,400 grants from the National Endowment for the Humanities. More than $100 million in congressionally appropriated funds awarded to scholars, writers, archivists, and researchers across the country was snatched up in three days. There weren’t any individualized reviews or hearings. There was no due process. Just a chatbot and two guys from DOGE who had no legal authority to be there in the first place.
Earlier this month, U.S. District Judge Colleen McMahon handed down a explaining exactly how illegal that was. She found a smorgasbord of constitutional problems. The mass cancellations violated the First Amendment because they were based on viewpoint discrimination. There were violations of equal protection because certain groups were systematically singled out for adverse treatment. Then there was the “ultra vires problem”: the DOGE officials who ran the whole operation had no legal authority to touch an NEH grant in the first place.
Complaints about the so-called deep state, a term favored by this administration and its supporters, have always centered on a specific grievance: unelected, unaccountable bureaucrats insulated by their own ideas about expertise and merit. The record assembled under oath describes, with uncomfortable precision, the administration doing exactly what it accuses everyone else of doing. Merit displaced by politics. Unaccountable actors. Ideological discrimination. Racial discrimination. The charges fit. It’s just not the defendants they had in mind.
The National Endowment for the Humanities was established by Congress in 1965 to fund scholarly and cultural work across the country. The statute that created the agency was clear about what Congress was funding: work of “substantial scholarly and cultural significance” that reflects “the diversity and richness of our American cultural heritage.” For those who believe that statutory text means what it says, it is worth pausing on that last phrase. Diversity was not smuggled into this agency by ideologues. It was written into the law by Congress.
To protect that mission, it built a layered review process. Applications go first to peer experts, who evaluate scholarly merit. Program officers with field knowledge make recommendations. The National Council on the Humanities—a body of 26 subject-matter experts appointed by the president with Senate confirmation—weighs in before the NEH chair makes a final decision. By the time a grant is approved, a lot of people who know what they are talking about have said so.
Enter Nate Cavanaugh and Justin Fox. These were the DOGE employees who canceled the grants. Here is the court’s language, not mine:
Prior to joining the Trump Administration, neither Fox nor Cavanaugh had any experience in government, public grant administration, private grant administration, or reviewing humanities projects for scholarly merit. In fact, as both were in their twenties, they did not have much experience in anything at all—certainly not in anything remotely related to the humanities.
Bad enough. But their process for deciding which grants to cancel—allotments that determined whether scholars could pay their rent, take their sabbaticals, and finish their books—made their lack of experience look like the least of their problems.
NEH staff produced a spreadsheet flagging Biden-era grants for perceived association with DEI, environmental justice, or “gender ideology.” Fox also had his own spreadsheet, built around a Control-F search for terms like tribal, immigrants, diversity, inclusion, Indigenous, Native, equity, equality, marginalized, BIPOC, solidarity, citizenship, melting pot, social justice, and gay.
What’s worse, Fox and Cavanaugh did not even read the underlying applications before issuing cancellations. They didn’t consult project websites or other supplementary materials. Spreadsheet-buzzword descriptions were enough for them. They also did not seem to read or care about the statute that governs the agency and states that these exact subjects are proper objects of NEH support. The law’s mandate was the diversity and richness of America’s cultural heritage, but DOGE’s methodology was Control-F.
Two guys in their 20s with no relevant experience used a chatbot to terminate the grants of scholars who had spent careers earning the kind of expertise the process was designed to evaluate. The administration claiming to have come to Washington to restore merit and accountability sent unqualified people to do a job they had no authority to perform. To make matters worse, government lawyers later argued that there was no constitutional problem because the decisions were ChatGPT’s doing, not the government’s. The chatbot did it.
Next up is viewpoint discrimination. The administration has spent the past year and a half arguing that federal funding was being used to promote an ideological agenda that elevated one set of views at the expense of others. There were actually two issues here.
First, only Biden-era grants were reviewed for termination. That was the big filter. Grants awarded during the first Trump administration were not examined at all. The ideological scrutiny was applied exclusively to one administration’s grants. As the court found, these grants were penalized not because of what they said but because of when they were awarded.
Notwithstanding Fox and Cavanaugh’s failure to define DEI in their “ChatGPT-derived methodology,” the court did the job for them. McMahon relied on dictionary definitions, one of which describes DEI as “the idea that people have equal rights and treatment and be welcomed and included.”
She also framed DEI as the viewpoint “that the exclusion of historically disadvantaged groups is undesirable.” The government canceled grants that expressed that view, and it didn’t hide it. On the spreadsheets, each grant carried a “Yes/No DEI?” column and a “DEI Rationale” column. That rationale column did not apply neutral statutory criteria; in some cases, the flagged grants had little to do with DEI at all. (I’m thinking here of a project that sought to recover ancient writings attributed to Moses and was classified as DEI because it claimed to “provide important insight into Jewish thought from two thousand years ago.”) As McMahon put it, asking whether a grant “relates at all to DEI” reflects a purely ideological filter with no other purpose than to identify disfavored perspectives the government wants to suppress.
This is against a backdrop in which universities have been pushed to demonstrate their commitment to “viewpoint diversity” or risk losing precious federal dollars. Indeed, some institutions have eliminated whole offices, revised curricula, and restructured programs. But according to this federal court, the administration has itself engaged in textbook viewpoint discrimination.
Finally, there were equal protection problems tied to identity-based discrimination. Fox’s spreadsheet flagged grants containing words like BIPOC, minorities, Native, tribal, Indigenous, and immigrant. As a result, the ChatGPT-generated DEI rationales classified grants as DEI precisely because they involved specific races, ethnicities, religions, or genders. Protected characteristics were functioning as the criteria for termination.
The pattern in the outcomes is telling. The government terminated grants involving countries in Africa, Asia, and South America. But it retained grants involving Western European subjects. Projects on “Victorian poets,” a “German theologian,” a “German philosopher,” “British philosophers,” an “English philosopher and mathematician,” and an “Italian composer” were worthy of preservation. Meanwhile, projects were deemed wasteful “because they related to Blacks, women, Jews, Asian Americans, and Indigenous people.” McMahon found that this amounted to discrimination based on race and that the terminations were unlawful.
What that means in practice is less clear than it sounds. The opinion explicitly stops short of ordering the government to pay out the canceled funds or resolving any individual grantee’s contractual claims. For the scholars whose awards were canceled before they ever started—who had already reorganized their lives, turned down other work, and made irreversible professional decisions in reliance on funding that never arrived—the legal victory and the practical reality are different things. The government will likely appeal, and we will see how long this ruling holds. If it holds, the terminations are void and the government cannot reimpose them. But recovering the money would require separate proceedings—a reminder that winning a case and recovering what you lost are not always the same thing.
We are in a moment when the institutions that produce and transmit knowledge—K–12 schools, universities, archives, and media outlets—are under sustained pressure. The NEH case is one data point in that larger story, but it is an unusually well-documented one. Because it went to litigation, because people testified under oath, and because a federal judge assembled the record, we know exactly what happened. Unaccountable actors. No merit. Unconstitutional from start to finish. In 143 pages, a federal judge documented what it looks like when the people diagnosing the disease are the ones actually spreading it.
