A group of scientists filed a class-action complaint on 16 September 2026 against the National Institutes of Health, alleging that the agency screens both new applications and existing awards for language associated with viewpoints the current administration dislikes. The American Civil Liberties Union, which is representing the plaintiffs along with the ACLU of Northern California, Protect Democracy, the Center for Science in the Public Interest and Jacobson Lawyers Group, said the case was filed in San Francisco and asks for class-wide relief covering researchers whose awards were terminated as well as those facing denials, delays or forced renegotiation.
What the complaint actually alleges
According to the ACLU’s own account of the filing, the NIH has been using a list of 235 keywords as proxies for disfavoured research, including “equity”, “advancing diversity”, “racist”, “gender”, “Latinx”, “climate change” and “pregnant individual”. The organisation says some grants and applications are terminated or denied outright, while in other cases researchers are asked to strip flagged words from their materials before the file can move forward, and a refusal ends the process. The plaintiffs argue that this layer of review sits on top of applications that have already cleared scientific peer review and been endorsed by the officials Congress designated as the decision-makers, and that it therefore breaches the First Amendment as well as statutes and regulations governing how public health grants are awarded and cancelled.
Mainstream coverage carried the allegations quickly. The Guardian reported that the researchers argue the Trump administration is blocking and cancelling scientific grants to censor topics with which it disagrees. UPI reported that the researchers say the blocking and cancelling of grants on disliked topics amounts to unconstitutional censorship, keeping the characterisation in the plaintiffs’ mouths. None of these accounts, at the time of filing, included a substantive NIH response on the record.
Whether courts can police grant-making
The uncomfortable part of this case is that federal science funding has never been a neutral machine. Congress appropriates money for stated purposes, institutes publish strategic priorities, study sections apply judgement about what counts as significant, and every administration tilts the portfolio towards the questions it considers urgent. A researcher whose proposal loses on those grounds has always had the same complaint available in principle, which is that the state decided some questions were worth public money and others were not. The plaintiffs’ stronger claim is narrower and more interesting, namely that a word-matching filter applied after peer review is not a priority judgement at all, because it screens by vocabulary and by the identity of the applicant instead of by the scientific quality of the proposal. That is a claim a court can examine against the statutory text, and it is a very different argument from a general objection to political influence over science.
It is also fair to ask whether the same plaintiffs and the same advocacy organisations would be in federal court if the keyword list ran in the other direction and screened out proposals that failed to address named social priorities. The complaint, as described by the ACLU, does not ask for a funding process insulated from policy choices in general. It asks for the restoration of the specific statutory arrangement that existed before, under which peer reviewers and named NIH officials made the call. Whichever way the litigation goes, the case will produce a public record of how much ideological filtering the process can legally absorb, and that record will be available to whoever holds the agency next.
The department’s own framing of its scientific standards is worth weighing against its recent conduct. Ars Technica reported in July that an FDA advisory committee reshaped by Health and Human Services Secretary Robert F. Kennedy Jr. voted to lift restrictions on the manufacture of several peptides, with the members added in June supplying the decisive votes and no clinical safety or efficacy evidence presented at the meeting, despite Kennedy’s earlier statement that independent experts would rigorously evaluate each substance using full clinical, pharmacological and safety evidence. Kennedy was confirmed as secretary in February 2025 by a Senate vote of 52 to 48, according to the public record of his confirmation, and FactCheck.org maintains a running file of fact-checks on his public statements about vaccines and public health. An agency that invokes rigour in one proceeding and applies a vocabulary filter in another invites exactly the kind of scrutiny this lawsuit is designed to force.
How the outlets framed it
The Guardian led with the accusation itself, headlining a lawsuit that accuses the Trump administration of blocking NIH grants as a form of censorship and foregrounding the ACLU’s role as counsel, which places the reader inside the plaintiffs’ theory of the case from the first line. UPI used cooler construction, reporting that researchers sued over censorship claims and that they say the policies amount to unconstitutional censorship, which leaves the characterisation as a contested allegation. The ACLU’s own press release goes further than either outlet by asserting the keyword list and the constitutional violation as settled fact, which is what an advocacy filing is meant to do. The difference matters because the legal question is not whether the screening is unpopular among scientists but whether it breaches specific statutes, and the more assertive framing quietly treats that unresolved question as already answered.
