Climate scientists who depend on National Science Foundation grants are quietly rewriting their proposals, swapping terms like “equity,” “diversity,” and “inclusion” for less politically charged alternatives. The shift follows a Senate investigation that flagged $2 billion in NSF awards as “woke DEI grants” and a January 2025 executive order directing federal agencies to end DEI-related programs. Researchers now face a practical dilemma: use language that accurately describes their work and risk losing funding, or adopt coded substitutes that satisfy new political filters but obscure the science.
The $2 Billion Target List
The pressure campaign against DEI-linked science funding became concrete earlier this year when the Senate Commerce Committee, chaired by Sen. Ted Cruz (R-Texas), published a searchable database of NSF awards it characterized as tied to DEI initiatives. In its announcement, the committee said it had identified roughly $2 billion in grants it viewed as problematic, giving critics of the agency a ready-made list for further scrutiny and future hearings. That release came weeks after the White House issued an executive order ending federal DEI programs, instructing agencies to inventory and, where appropriate, terminate or reshape grants that advanced diversity-related goals.
Together, these actions created a two-front squeeze on researchers. The executive order set the policy direction; the Senate database named specific targets. For climate scientists, whose work frequently involves studying how environmental hazards affect different communities unevenly, the overlap between standard scientific vocabulary and politically flagged terminology is especially wide. Words like “underserved,” “marginalized,” and even “community” can appear in both a climate-vulnerability study and a DEI initiative, making climate proposals particularly exposed to keyword-based reviews.
NSF’s Keyword Contradiction
The NSF’s own guidance creates a confusing signal for grant applicants. On a public page explaining how it is implementing recent executive orders, the agency states that it “does not recommend using a keyword-based approach” for proposal compliance and emphasizes that it will instead conduct broader portfolio reviews in light of the new directives. Yet internal practices have pointed in a different direction. Reporting from The Washington Post described staff combing through projects using a keyword list, with terms flagged differently depending on whether their usage was deemed DEI-related or not, and an internal flowchart guiding this triage.
That gap between official advice and internal practice is what drives the self-censorship. Researchers cannot rely on the agency’s assurance that keywords alone will not trigger problems when evidence suggests keywords are exactly how the review process works behind closed doors. From a grant-writing standpoint, the rational response is to avoid any term that might trip a filter, regardless of whether it accurately describes the research questions or the affected populations.
The Language Workaround Spreads
The pattern extends well beyond climate science and the NSF. According to reporting in the health-research community, hundreds of projects funded by the National Institutes of Health were modified in the first fiscal year after the executive order to scrub diversity-related language from their descriptions. At the Department of Health and Human Services, the administration went further: officials barred Head Start providers from using words like “women” and “race” in grant applications, according to the Associated Press. A federal judge later blocked that particular restriction after a list of discouraged words surfaced in court filings, but the episode showed how quickly informal guidance can harden into enforceable rules.
The Head Start case is instructive because it traces a full arc, quiet word lists, formal directives, and then legal challenge. Climate researchers watching that sequence have reason to believe that today’s informal keyword screening at the NSF could become tomorrow’s explicit prohibition, especially if future reviews of the Senate’s database of “woke” grants lead to more aggressive oversight.
New Tools for a New Compliance Reality
The demand for sanitized grant language has already produced its own cottage industry. A peer-reviewed paper published in late 2025 described an AI-assisted compliance platform, dubbed GrantCheck, designed to help researchers align their narratives with current policy requirements while “maintaining scientific innovation.” The tool analyzes text for potentially sensitive terms and suggests alternative phrasing that it predicts will be less likely to raise red flags with federal program officers or political appointees.
Its existence signals that the language-swapping behavior is not an anecdotal quirk but a recognized, systematic need. In interviews, climate scientists describe informal spreadsheets and shared glossaries that serve a similar purpose: “vulnerability” becomes “exposure,” “equity” becomes “distributional impacts,” and “communities of color” becomes “geographically clustered populations.” The scientific intent remains, but the social context is flattened.
At the same time, the NSF has relaxed its own merit-review requirements, reducing the number of outside reviews needed for proposals in an effort to address a growing backlog and staffing constraints. Fewer reviewers per proposal could mean less expert judgment about whether a reworded study still makes scientific sense, or it could mean fewer opportunities for a flagged keyword to draw attention. Either way, the review process is changing on multiple fronts simultaneously, and researchers are adjusting their writing to match a moving target.
Grant Terminations and the Five-Word Formula
The stakes of getting the language wrong are not abstract. The phrase “no longer effectuates agency priorities” has become the standard justification for terminating existing awards in response to the new executive order, according to internal correspondence and cancellation notices obtained by reporters. In one investigation, The Washington Post documented how that five-word formula appeared in letters cutting off funding for projects ranging from climate resilience to social-science research, often with little additional explanation.
For climate scientists, that phrase is especially chilling because “agency priorities” are now partially defined through the lens of the Senate’s DEI database and the White House order. A study framed as protecting “frontline communities” from extreme heat might once have been celebrated as responsive to federal goals; today, the same language can be cited as grounds for cancellation. Researchers have begun quietly revising the public abstracts of ongoing grants, not just new proposals, to minimize the chance that a stray word will draw scrutiny.
Transparency Battles and FOIA Fights
As grants are reworded or revoked, outside groups and journalists have turned to transparency laws to understand what is happening inside the agency. The NSF’s own Freedom of Information Act policy explains how members of the public can request records about funded projects, internal guidance, and communications with political appointees. But FOIA requests can take months or years to resolve, and heavily redacted responses often leave key questions unanswered.
Advocacy organizations have filed requests seeking the full keyword lists used in internal screenings, along with any instructions sent to program officers about implementing the executive order. So far, only fragments have emerged, often through litigation. That limited visibility leaves researchers guessing about which terms are safe, which are risky, and how those judgments might shift with each new round of political pressure.
Climate Science in Code
The cumulative effect is a kind of coded writing that obscures who is most affected by climate change. Instead of explicitly studying how sea-level rise threatens Black neighborhoods in a coastal city, a proposal might now refer to “historically developed low-lying census tracts.” Instead of examining wildfire smoke exposure in farmworker communities, it might describe “outdoor labor populations in high-burn regions.” The underlying science (modeling heat waves, mapping floodplains, tracking air quality) remains rigorous, but the social meaning is blurred.
Some researchers worry that this shift will eventually affect what gets studied at all. Questions that cannot be easily rephrased without losing their core (such as how racism shapes disaster recovery or how immigration status affects evacuation) may simply fall off the agenda. Early-career scientists, whose jobs depend on securing grants, are especially likely to steer away from politically sensitive topics, even if they can technically be framed in neutral language.
What Comes Next
For now, climate scientists are navigating a narrow channel between political directives and scientific integrity. They are learning to write for two audiences at once: reviewers who understand the coded language and political overseers who may scan only for disfavored words. The result is a body of work that often says less, on paper, than researchers intend, and that may be harder for policymakers and the public to interpret accurately.
How long that arrangement can last depends on forces largely outside the scientific community’s control: future court rulings on the executive order, congressional oversight of the NSF, and the willingness of agency leaders to defend research that addresses unequal climate impacts in plain language. Until then, the country’s primary funder of basic science remains caught between its mandate to advance knowledge and a political environment in which a single word can determine whether a climate study lives or dies.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
