Science1 publisher3 min readPublished
Seventeen researchers sue to stop NIH from screening grant applications for keywords
The complaint was filed as a class action, so a ruling for the plaintiffs would reach every applicant the term searches touched. NIH's own head of extramural research has already described the practice in a court filing.
The Scientist · Science desk
What happened
- Thirteen named researchers and four anonymous ones sued NIH, asking a court to stop the agency from searching applications and current grants for specific terms to decide whether to reject them.
- The complaint was filed as a class action, so a decision for the plaintiffs would apply to any researcher whose work the term searches affected, not only the 17 scientists who sued.
- The screening has largely focused on grants in areas the Trump administration disfavors, including work to diversify the research workforce and studies of health disparities, climate change and vaccines.
- NIH moved from mass terminations to scrutinizing the language of the grants it funds, with officials asking grantees to tweak wording in a process that NIH employees and grantees call renegotiating.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- precedent A ruling on speech grounds would go at the reason an application was rejected, not the paperwork behind it; the ACLU's Olga Akselrod says the vacated directives were replaced by new ones that continued the same purge.
- exposure Researchers who never sued are inside the case if the searches touched their work. That makes NIH's own flagging records the measure of how big the affected group is.
- constraint When search terms govern eligibility, an applicant has to establish standing in a field while avoiding the vocabulary that names it, and that pressure operates on new proposals no court will ever read.
NIH director Jay Bhattacharya told the conservative advocacy group PragerU in July, "I am deeply against banned word lists. I don't believe in them. I will not have my institution embrace them" [14]. In a stipulation filed that same month, in a case brought by University of California researchers fighting terminations aimed at their institution, the agency's head of extramural research wrote that "NIH used search terms, keywords, or phrases to identify and terminate grants presumed to express viewpoints ... that HHS-NIH no longer wished to subsidize" [15]. Bhattacharya has repeatedly denied that his agency screens for forbidden phrases [13]. So the plaintiffs start from inside the record, with the agency's own filing saying the searches happened.
The new suit argues speech, not procedure. The first round of terminations came undone on procedural grounds: a federal judge ruled that the slap-dash nature of the terminations violated the Administrative Procedures Act, and hundreds of grants were reinstated [5]. Guidance can be rewritten to satisfy the APA, and other suits built on that statute have stalled [6]. "NIH supplemented the directives at issue in APHA with a series of additional directives, guidance, and policies that purported to replace the vacated directives, but in fact just continued the viewpoint-based purge of existing grants. It's more of the same," said Olga Akselrod, a senior counsel at the ACLU who represented researchers in both cases [10].
Arghavan Salles, one of the named plaintiffs, had an R01 terminated. Her project was on the sexual harassment researchers experience during their training, and it was not reinstated [11]. "I couldn't talk about my research on implicit bias or microaggressions or sexual harassment or belonging. I couldn't use any of those words. It's really hard to demonstrate that you're a leader in a field without being able to say any of the words that describe that field and that work," she said [12]. Michael Green, a social epidemiologist at Johns Hopkins, applied in 2023 for a grant to carry him from his PhD into postdoctoral work on the health effects of discrimination in medical settings, when that work was being praised; last year, moving from the PhD portion to the postdoctoral portion, he was asked to remove certain words [17][18].
The public record is thin on scale. The one count in it is the more than 700 grantees who changed their grant titles in 2025 to remove phrases like "equity," "disparity," and "racism" [8]. That number tells you how many people complied. It doesn't tell you how many applications were flagged. STAT reports that the portfolio review was helped along by a "computational text analysis" tool [9]. NIH did not respond to STAT's request for comment on the lawsuit [16].
For now, these are allegations in a complaint. The scrutiny has also been applied to people applying for new grants [19], and if a judge rules for the plaintiffs, the reach of that ruling will be set by whichever of those applications the searches touched [3].
What to watch
- Whether a judge certifies the class, and how narrowly the covered group of applicants is defined.
- Whether discovery produces the term list or the outputs of the computational text analysis tool.
- How NIH's answer to the complaint squares Bhattacharya's public denial with the July stipulation.