A few months ago, Sen. Ted Cruz announced that he had uncovered $2 billion of science grants funded by former President Joe Biden’s administration that prioritized “radical political perspectives” or “neo-Marxist theories.’’ His aides on a congressional committee assembled the list by searching the project descriptions for 699 key terms like “women,” “diversify,” “segregation” and “Hispanic culture.”

When Cruz released the database of this allegedly “woke” research earlier this month, we decided to run our own experiment. We asked one of the models powering ChatGPT, which can sift through large amounts of data, to evaluate all 3,500 grant descriptions in the database as if it were an investigative journalist looking for Marxist propaganda, “woke ideology,” or diversity, equity and inclusion. The model tried to give us descriptions of how each project might fit those themes. We were particularly interested in the grants where it came up blank. We then read through the researchers’ full summaries of those and many other grants, including each one described in this story, looking for references to some of the keywords on the list.

We found that Cruz’s dragnet had swept up numerous examples of scientific projects funded by the National Science Foundation that simply acknowledged social inequalities or were completely unrelated to the social or economic themes cited by his committee.

Among them, for example, was a $470,000 grant to study the evolution of mint plants and how they spread across continents. As best we can tell, the project ran into trouble with Republicans on the Senate Committee on Commerce, Science and Transportation because of two specific words used in its application to the NSF: “diversify,” referring to the biodiversity of plants, and “female,” where the application noted how the project would support a young female scientist on the research team.

Other projects our AI assistant led us to included:

• Developing a device that could treat severe bleeding. It seems to have caught the committee’s attention for using the words “victims” — as in gunshot victims — and “trauma.”
• Creating biosensors to detect infectious diseases. The grant appears to have been tagged for the repeated use of “POC,” an acronym often used for “people of color” but in this context meaning “point of care” — that is, the place where people receive medical treatment — and “barrier,” referring to a part of the biosensor itself.
• Designing eye-tracking technology for diagnosing and treating concussions. It appears to have gotten flagged for referencing “traumatic” brain injuries and the “status,” meaning the condition, of patients.

It’s “very frightening,” said Charlotte Lindqvist, a biology professor at the University at Buffalo who is conducting the research on mint plants.

Lindqvist spends hours a day grinding up plant samples and analyzing their DNA to identify genetic differences between species. Studying plant diversity, she said, could help secure more resilient food systems. “We are really trying very, very hard … to move our world forward, understanding it better through our sort of foundational, sometimes groundbreaking research,” she said, “and then you get flagged and blacklisted because there is a word like ‘female’ in your project.”

Staff for the Republicans on the Senate committee assembled their report by examining all NSF grants awarded to projects that began between January 2021 and April 2024. Using their list of keywords, they flagged those earmarked for research that they said was “often based on neo-Marxist theories that identified merit by physical or ethnic attributes, not one’s talent, work ethic, or intellectual curiosity.”

Evaluating the merits of these awards would require a deep understanding of dozens of scientific fields, from gravitational waves to DNA methylation. But the report describes a crude approach; while staffers did attempt to account for the different ways their keywords can be used, they did not manually review all grants. The report also failed to acknowledge that the NSF has a legal mandate to make science more inclusive of women, racial minorities and disabled people.

Cruz released the full database just as the Trump administration’s NSF said it was examining research grants to make sure they complied with the president’s executive orders terminating diversity, equity and inclusion initiatives. Cruz said he requested “significant scrutiny” of the grants in his database. At the time, the NSF was using a similar list of keywords for its review.

Neither Cruz’s office nor a spokesperson for Republicans on the committee responded to requests for comment.

It’s not clear if approved projects that are still waiting for payments will get their money. A federal judge ruled last Friday that the administration can’t cancel or freeze grants for supporting diversity, equity and inclusion programs. When asked how it would respond to the judge’s preliminary injunction, an NSF spokesperson directed ProPublica to an agency webpage, which had not been updated with information about the court ruling at the time of publication.

“NSF is working expeditiously to conduct a comprehensive review of our projects, programs and activities to be compliant with the existing executive orders,” a spokesperson told ProPublica in response to questions about its review process.

The Senate committee’s list includes words like “diversify” and “biases,” which have technical meanings unrelated to social issues. Although the report’s authors worked to remove grants flagged for those reasons, some, like Lindqvist’s, slipped through.

The lack of precision in the committee’s methodology is “obviously laughable,” said Kim Lane Scheppele, a professor of international affairs at Princeton University who studies the rise and fall of constitutional governments. But she also worries about what might happen if lawmakers take a more serious approach, such as trying to ban research on racial inequality, similar to how Congress severely limited studies on gun violence.

The NSF evaluates grant proposals based on two factors. The first is intellectual merit. Every application is reviewed by a panel of experts — often other academics — who specialize in the same topic. They pore over detailed applications that include data, references and researchers’ qualifications, far more information than the brief summaries evaluated by the Senate committee.

The other factor is “broader impacts,” which could include how the research might benefit societal well-being or make science more inclusive.

Currently, federal laws require the NSF to support research at historically Black colleges and universities and other institutions that serve groups who are underrepresented in science. Congress also ordered the NSF to fund efforts “designed to increase the recruitment, retention, and advancement” of members of these groups in scientific careers.

“All of that is hard-wired into federal funding,” Scheppele said. “If anyone was ‘woke,’ it was Congress.”

Laws passed by Congress have more legal weight than executive orders, so the NSF shouldn’t prioritize Trump’s order over its mandate to support underrepresented people in science, Scheppele said. The White House, she said in an email, is “literally asking the NSF to violate the law!”

The committee report singled out some projects for simply acknowledging that people from certain demographics face unique challenges. That includes a University of Houston study of maternal mortality that examines why Black, Indigenous and other people of color in the U.S. are nearly three times as likely as white women to die during pregnancy or within the first year after childbirth. Another project, which involved using drones to deliver defibrillators to people suffering cardiac arrest, appeared to be flagged because it noted that emergency response times are slower in low-income and minority neighborhoods.

In other cases, the keywords that caught the committee’s attention may have come from outreach efforts meant to broaden the impact of the research. A $6 million nuclear astrophysics project to study the origins of the universe includes a reference to attracting a “diverse group” of students interested in the subject and a summer school program for increasing interest in nuclear-science careers, “especially among women and minorities.”

That’s in line with a 1998 law that ordered the NSF to develop “intellectual capital, both people and ideas, with particular emphasis on groups and regions that traditionally have not participated fully in science, mathematics, and engineering.”

Congress recognized “you’re going to get better science” that way, said Melissa Finucane, vice president of science and innovation at the Union of Concerned Scientists. When you get different perspectives interacting and thinking about complex problems, she said, you’ll get different and new ways of solving a problem.

The report’s “sledgehammer” methodology ignores the substantial scientific merit of these projects, many of which address “critical national needs in areas such as aerospace, agriculture, and computing infrastructure — as well as the need to broaden the talent pool,” a spokesperson for Democrats on the Senate committee said in an email. The email said that ranking Democrat Sen. Maria Cantwell of Washington “understands that there is no way the United States can compete” with the rest of the world on innovation “without ensuring that NSF funding emphasizes the participation of women and minorities in STEM,” a reference to science, technology, engineering and mathematics.

Rice University professor Vicky Yao has seen firsthand how efforts to broaden participation can increase excitement and interest in science.

When Yao applied for a research grant in 2022, she included outreach to community college students, many of whom are from underrepresented populations and don’t have access to research opportunities.

When ProPublica informed Yao her $610,000 project was on the Senate committee’s list, she found it bizarre that such technical work on DNA methylation — a process that can affect cancer and neurological diseases — could be labeled as “woke.”

The committee’s choice of keywords is so sweeping that shutting down the research that uses those terms would end not just diversity programs but also vast fields of research on social science (“Black communities,” “racial inequality,” “LGBT”), climate change (“net zero,” “climate research,” “clean energy”) and medicine (“white women,” “victims,” “trauma”).

If any research related to women or minority populations is under fire, then “we’re talking about maybe 65% of the American population. So at that point, what’s left?” said Dominic Boyer, an anthropology professor at Rice University whose project on reducing flood risk was flagged by the committee. “Under what authority, or according to what philosophy, can a government invalidate or discredit research that’s focusing on two-thirds of the population?”

Boyer received an award of $750,000 to use nature-based solutions like rain gardens to reduce flooding in Houston, where Hurricane Harvey displaced tens of thousands of people in 2017. His team has begun collaborating closely with residents from three neighborhoods: two lower-income communities where the residents are mostly Hispanic, Black or Asian, and a middle-income neighborhood with mostly Hispanic and white residents.

He initially assumed that’s why his research was flagged. But it turned out that the triggering keywords may have come from boilerplate language that describes the specific NSF program that funded Boyer’s work: Strengthening American Infrastructure. The portions of the grant’s program description containing those keywords were written by the NSF during Trump’s first term. It used the words “socioeconomic” and “equal opportunity” to explain why infrastructure is important to society. The same description is found in more than two dozen other grants on the committee’s list.

Boyer said it speaks to a kind of “Orwellian absurdity” that “these words can only have one meaning, and it’s the meaning that they would like to politicize.”