The term “chilling effect” has come up frequently during the Trump 2.0-era philanthrosphere, as some funders have backed away from supporting fields like DEI, climate change and democracy because they’re afraid — justifiably, as it turns out — of raising the administration’s ire.
The midterm election results have diminished President Donald Trump’s power, but they haven’t allayed those fears. Progressive leaders understand that the administration still plans to reshape parts of the charitable sector, and if recent Wall Street Journal reporting is any indication, these efforts may still include efforts to revoke some groups’ tax-exempt status, although it’s not yet clear which organizations will be targeted.
A new report from the nonpartisan Coalition for Independent Technology Research (CITR), which seeks to advance, defend and sustain the right to ethically study the impacts of technology on society, points to a similar “chilling effect” in tech research.
CITR used “chilling effect” to describe how the administration is using levers of power, including the threat of deportation, to dissuade researchers from digging into areas like disinformation, internet governance and sociotechnical questions around profiling or discrimination. Coalition members also report that some funders are dialing back support for politically sensitive technology research, particularly disinformation work.
This isn’t to suggest money isn’t flowing to researchers. The problem, the report argues, is that a growing share of support comes from “Silicon Valley-aligned funders” bankrolling “studies on speculative and mythical future harms” instead of urgent, short-term threats. CITR also makes the case that tech companies are funding studies that advance their own agendas while shutting out independent voices that could portray them in an unfavorable light. As a result, 72% of surveyed independent AI researchers cite “lack of transparency from AI Developers” as a major barrier to their work.
Funders’ broader efforts to push for AI regulation and support safety work have always been tempered by the stark reality that, by being financially outmatched by tech companies and operating in an environment where the federal government under Trump hasn’t shown much interest in robust federal AI oversight, they can only move the needle so far. Now, CITR’s research claims some funders are dialing back support for critical research areas because they don’t want to antagonize the administration or tech companies, leaving those who remain to consider whether — and how — to fill the gaps.
Funders have dialed back disinformation research
Titled “State of Independent Technology Research 2026: Resisting Corporate Capture,” CITR’s report discusses how coalition members — academics, journalists, civil society researchers and community scientists — have reported that funders are especially hesitant to support AI-related disinformation research.
For instance, a member reported that their previous disinformation work made them “too politically risky to fund in the current environment,” according to one funder. Another coalition member reported that funders “are scared they / their grantees will be sued,” while a third was advised to change the wording on their website to avoid drawing attention.
The chilling effect here has its roots in the early months of the second Trump administration. In April 2025, Elon Musk’s DOGE announced that the National Science Foundation (NSF) had canceled at least 402 grants. The cuts affected awards “including but not limited to those on diversity, equity and inclusion (DEI) and misinformation/disinformation,” wrote Nieman Lab’s Sarah Scire. Two canceled grants related to disinformation sought to develop AI-based techniques that would give social media users robust context about the posts they see online.
Scroll through the names of terminated Department of Health and Human Services grants, and you’ll find a handful of recurring words, like “equity,” “HIV/AIDS,” “vaccines” and “diverse,” as if a DOGE staffer ran a search on the department’s grants database and terminated those with the offending terms.
“The focus on disinformation research specifically — and in particular, shying away from using the word ‘disinformation’ — directly aligns with the Trump administration and DOGE’s approach of creating a ‘blacklist’ of certain terms that render research projects ineligible for funding,” said CITR Executive Director Brandi Geurkink, via email. “When philanthropy takes this position, it cedes ground and prevents important research from happening.”
Nonprofit leaders have spent the better part of the past two years confronting the fact that some funders, gripped by the Trump-administration-induced chilling effect, have withdrawn support for related hot-button issues like racial equity, LGBTQ rights and climate change.
The good news is that, according to Center for Effective Philanthropy President Phil Buchanan, this phenomenon doesn’t appear to be widespread throughout the sector. In a separate conversation on how foundations are adjusting their Trump 2.0-era due diligence processes, he told me, “I know of many other foundations that have maintained their work in these areas or doubled down.”
Geurkink shared a similar takeaway. “Not all funders are doing this,” she said, alluding to a retreat from supporting important tech research. “We have seen several examples of funders changing the way that they fund, and what they require organizations to do to receive that funding, in response to the risk environment.”
Tech-aligned entities are making enormous funding commitments
Of course, plenty of entities are cutting checks for AI research. The catch, according to CITR, is that many of them are closely connected to Silicon Valley.
A CITR coalition member is currently working on an analysis looking into the flood of funding coming in from industry and industry-aligned charities, which, Geurkink argued, tend to be concentrated among industry-preferred researchers (Geurkink cited the prominent grantmaker and philanthropic advisor Coefficient Giving as one such funder). The member “has identified $1 to 2 billion so far, mostly over the past couple of years, from those sources alone,” she said. One commitment from Nvidia — a noncash “compute donation” valued at $250 million over five years to support U.K. agentic safety and security research — “exceeds a typical year of the NSF’s Directorate for Social, Behavioral and Economic Sciences funding in a single grant.”
In July, Anthropic announced that its $200 million Economic Futures Research Fund would focus on “external research on interventions to prepare society for the economic impacts of AI.” “Here too, an industry funding initiative exceeds the NSF’s entire funding of behavioral research last year,” Geurkink said. “These massive influxes of industry money are unprecedented and, to our knowledge, have no parallel in the history of science.”
Geurkink also noted that research labs and centers receive significant funding or data access from the companies they study. “The concern here is that when resources concentrate in this manner, you run the risk of creating feedback loops,” said the CITR report’s author, Vineet John Samuel, a researcher studying the economic and social impacts of social media and AI platforms on developing countries. “In this scenario, industry-backed researchers win grants over independent scientists, end up shaping the research agenda, and move into roles like peer reviewer or editor where they can discourage critical inquiry into the industry.”
Around the same time CITR published its report, “EyesOnOpenAI,” the coalition of more than 50 California-based civic, labor, nonprofit and philanthropic organizations committed to holding OpenAI accountable for its nonprofit assets and governance structure, issued a Report Card grading the company in areas like Charitable Grantmaking (“D”), AI Safety (“F”) and Teen and Youth Safety (“F”). The AI giant’s AI Resilience grant program aims to “mitigate the risks of AI, so that society can maximize its benefits,” but as we’ve written before, conflict of interest concerns are numerous.
“When the attorneys general of California and Delaware signed off on OpenAI’s restructuring and laid out the memorandum of understanding, safety was at the center,” said EyesOnOpenAI coalition leader and TechEquity Collaborative cofounder and CEO Catherine Bracy. “It’s really the only thing of substance that they required OpenAI to commit to. In our view, they have not met that commitment, either in the spirit or the letter of the MOU, and so that’s why we gave them an F [in “AI Safety].
Scrutinizing the independence of some tech-funded AI research
Being deeply attuned to AI watchdogs’ safety concerns — to say nothing of the public’s growing suspicions of AI and data centers — tech companies go out of their way to stress that the research they fund is, in fact, independent.
Geurkink directed me to an Anthropic blog post about how the company is “enabling independent research” on how people use Claude. The post notes that researchers were “free to publish their results even if they are inconvenient for Anthropic.” To the casual observer, this all seems reasonable enough. Geurkink, however, isn’t sold.
“Unsurprisingly,” she said, “the research questions that the researchers chose — which Anthropic says they had no role in influencing — each reflect underlying hypotheses of the potential good that Claude could bring about in the world, and largely align with the picture the company wants to paint of itself externally.”
Geurkink went on to note that “the single ‘negative’ finding from the ‘independent’ studies, centered on people experiencing friction when ‘collaborating with AI,’ is largely one that only requires a simple technical fix, as opposed to the negatives that are structural or inherent within the operation of the technology itself.” Those negatives, Geurkink said, could span labor and human rights violations, copyright infringement, or environmental impacts stemming from the underlying build and deployment of the technology.
Are tech funders sufficiently bankrolling research into “present-day harms?”
Geurkink’s comments underscore another key theme running through the coalition’s report — AI funders’ unwillingness to vigorously back research tackling present-day harms, including criminal profiling and algorithmic bias, that could tarnish their brands, scare off investors, and raise concerns among regulators. CITR also found that small nonprofits researching AI’s harms are facing “aggressive subpoenas from big AI labs,” with 55% of surveyed researchers reporting “a worsening environment in their personal safety.”
The research that tech companies tend to back themselves often focuses elsewhere.
Anthropic, through the Economic Futures Research Fund, said it plans to look at “shaping AI’s impact on workers at the firm and workplace level” and “equipping people to navigate AI-driven transitions.” This may be an urgent priority, depending on the job in question, such as those in sectors most exposed to automation. However, to some, this work — which, it should be noted, is receiving substantial support from major philanthropic players — lacks the urgency of, for example, addressing data centers’ real-time environmental impacts.
Similarly, Coefficient Giving moves money through a pooled fund called “Navigating Transformative AI,” which seeks to “reduce the risk of global catastrophe and help society prepare for major advances in AI.” It’s valuable work (nobody wants a global catastrophe!), but the fund doesn’t explicitly address real-time AI harms laid out in the CITR report’s press release, like “endangering children’s safety,” “proliferating deepfakes,” and “generating misinformation.” Instead, the fund’s requests for proposals focuses on societal “capacity building,” “career development and transition funding” for individuals to pursue work “that could help to reduce global catastrophic risks,” and “technical AI safety.”
“The focus on supposed ‘existential risk’ draws much-needed resourcing away from understanding present-day harms that are being caused and perpetuated by the actions of AI corporations, while at the same time fueling the idea that the technological path we are on with AI is inevitable,” Geurkink said. “The midterm elections in the U.S. and the outright rejection of the billionaires’ agenda by voters demonstrate that it is not.”
Under normal conditions, private foundations could fill the gap by funding more independent research into AI’s real-world harms. But as CITR notes, some funders are looking over their shoulders because they’ve seen how the administration has made life incredibly difficult for their progressive peers. Added to the fact that they’re being vastly outspent by industry and industry-aligned entities, institutional philanthropies’ response to AI challenges leaves a lot to be desired.
That said, perhaps the recent election, in which many successful Democratic candidates campaigned for tougher oversight of AI and data centers, coupled with Trump’s impending lame-duck status, will compel funders to re-examine their reticence. Maybe we’ll look back in three years and find that they turned the spigot back on for hot-button areas like disinformation research.
Until then, Geurkink believes that the path forward for grantmakers is clear.
“If philanthropy wants to make sure that technological expansion benefits all of humanity, especially those most impacted by the social and political outcomes of this tech, they should support industry-independent researchers and scientists who are scrutinizing these systems, identifying safeguards, and imagining how AI could be built differently,” she said. “There are communities of people, such as our coalition, who have spent decades building methods for doing this work outside of industry.
“The problem is not that we don’t know how to do this research, or that we only need the companies to cooperate in order to do it. It is that incentives and capital flow toward capture and away from independence.”
Related Inside Philanthropy Resources:
For Subscribers Only
