Knowledge is a threat to authoritarian regimes insofar as it is accessible, viewed as legitimate, and independent from political control.48 The American federal government has always played a role in shaping the nation’s scientific priorities—its involvement is not new.49  However, a problem emerges when political authority extends beyond determining what research should be prioritized and begins to dictate what the evidence should say and who is allowed to say it. In other words, while the government has never been fully removed from the scientific process, it is nevertheless a core tenet of publicly-funded scientific research that, while public officials may direct priorities, political ideology should not dictate findings, suppress or erase inconvenient results, or determine the credibility of science or scientists according to political usefulness.

This independence from political ideology creates a particular problem for non-democratic administrations seeking greater control over institutions and the outputs produced by them. Research can document harms a government might prefer to minimize, contradict a politician’s preferred explanations, undermine the empirical basis of a politically preferred policy, or produce evidence supporting policies political leaders oppose. The threat posed by independent science to these regimes, then, is not necessarily what science says in any particular instance—it is that political power does not completely control what science is capable of saying.

The previous section examined this process structurally: the deconstruction and weaponization of the administrative state in pursuit of greater centralized control over science. In the previous section, we outlined the removal and restructuring of independent scientific review, the hollowing out of institutional scientific capacity, and the expansion of executive authority over scientific decision-making. But control over the institutions that produce science is only one dimension of control over science itself. Political influence can also operate over the production, circulation, and legitimacy of scientific knowledge—shaping what information remains accessible, which findings are treated as credible, what questions scientists are able to pursue, and who is recognized as a legitimate scientific authority.

In pursuit of greater political control over scientific narratives, governments may therefore seek not only greater centralized control over the production and circulation of research and scientific knowledge. In doing so it must also delegitimize science and scientists themselves. Across historical cases of political interference with science, several recurring mechanisms emerge. Here, we organize these mechanisms into five major pillars of the delegitimization of expertise and the scientific method: (1) removing access to information, (2) attacking scientists, (3) manufacturing uncertainty about established knowledge, (4) controlling what questions can be asked, and (5) replacing experts with loyalists. For each pillar, we describe the mechanism through a historical example, then examine how a similar mechanism has manifested under the second Trump administration.

Removing Access to Information

In July 1957, the Soviet Union’s State Censorship Committee compiled a list of topics that were not to appear in public reporting. Among the censored topics were forest fires, industrial and military accidents, radioactive contamination, and information on child mortality. The Soviet government was well aware these were problems facing the nation—experts, politicians, and even the Soviet secret service accumulated knowledge about the harmful environmental and human consequences of industrialization and environmental degradation. The real issue was that much of this accumulated knowledge did not fit within the state’s preferred account of Soviet progress. Information about environmental disasters and other consequences of industrialization were therefore restricted from public view.50 

However, evidence that contradicts a government’s preferred account poses little challenge if the public cannot see it and researchers cannot retrieve it. This constitutes our first pillar: removing access to information. In order for research to threaten political narratives, that research must be published, datasets must remain available, government reports accessible, and scientists, the public, and policymakers must all be able to retrieve the information. When politically undesirable knowledge becomes more difficult to access, the evidence itself does not require questioning—instead, alternative accounts to political narratives simply become less present in the information landscape, leaving more room for the government’s preferred account to dominate. 

In the modern era of technology, this does not simply manifest as book burning. Political restrictions can instead shape what information is published both in print and online, where that information is allowed to be published, and how easily it can be retrieved. For example, research examining more than 5,000 academic publications from scholars based at U.S. and Chinese universities found that under pressures associated with academic control and self-censorship in China, some politically sensitive topics effectively “disappeared” over time while others “migrated” from Chinese- to English-language publication venues.51 Control over access can therefore change not only what the public sees, but where knowledge is produced and circulated in the first place. 

 Since the start of the second Trump administration, substantial amounts of scientific information have disappeared from or been altered across federal websites. Climate information provides one particularly clear example. In June 2025, the federal website for the U.S. Global Change Research Program went offline, taking with it the primary federal portal through which the public could access the National Climate Assessments.52 These congressionally mandated, peer-reviewed assessments synthesize research on how climate change affects the United States, and had previously been publicly available through federal websites. While the administration initially indicated that the reports would be transferred over to NASA, NASA rejected this and publicly announced in July 2025 that it would not host them . 

The federal climate information infrastructure was further narrowed through changes to Climate.gov, a National Oceanic and Atmospheric Administration website.53  In 2025, the staff responsible for maintaining the website were terminated and the site stopped publishing new content altogether (ex-NOAA employees eventually came together to re-create the website in their personal time).54  Additional websites that have gone dark or otherwise been scrubbed under the second Trump administration include websites on topics including HIV, LGBTQ+ health and resources, vaccination guidance, civil rights information, gender-based violence resources and more.55 

Removing access is one way to diminish the influence of inconvenient evidence without successfully refuting that evidence. That said, governments need not necessarily remove information entirely to curb its influence. The evidence can remain entirely accessible while its credibility is attacked through another route we turn to now: undermining the scientists who produce the evidence. 

Attacking the Scientists

At the start of the twentieth century, a thriving community of geneticists existed in the Soviet Union. Among them was one Nikolai Vavilov, a prominent geneticist of the time, and a renowned plant scientist. Vavilov was passionate about traveling the world to collect diverse samples of plant seeds, and turned his work into an enormous collection of genetic plant material. By using Mendelian genetics56 and selective breeding, Vavilov was able to improve agriculture in the Soviet Union.57  In the 1930s, however, Soviet leadership began championing a competing geneticist with alternative ideas: Trofim Lysenko.58  Lysenko’s perspectives better mirrored the political ideology purported by the Soviet Union, rejecting the scientific consensus behind Mendelian genetics and the existence of genes in the first place, and arguing instead that plants could acquire characteristics from their environments and pass those on to future generations of plants.59  Thus, the pseudoscientific genetic ideology dubbed Lysenkoism60 was formed.

While this could have theoretically remained a scientific disagreement alone, it did not. Lysenko unsurprisingly found support for his ideas from Joseph Stalin himself, in addition to a larger body of Soviet political leadership. The dispute between Lysenkoism and Mendelian genetics increasingly moved beyond the bounds of scientific evidence, with Mendelian genetics recast as an ideologically corrupted science. The geneticists that supported Mendelian genetics were denounced as “reactionary,” their research as “bourgeois” and “anti-people.”61  Vavilov himself was arrested in 1940, dying in prison three years later, and other scientists who continued to defend Mendelian genetics were pushed out of their positions or otherwise prevented from working. 

An important lesson is to be learned from Lysenkoism: if you discredit the scientist on grounds of moral failings, you do not necessarily have to wade into the waters of actually disproving the science. This constitutes our second pillar: discrediting the scientists.

Most members of the general public, and politicians themselves, are typically not independently capable of replicating scientific studies, critically evaluating the results with an expertise perspective, or meaningfully deducing from competing interpretations of scientific findings. However, by simply raising into question the morality of a scientist, the scientific findings and deductions that scientist is championing become considerably less trustworthy in the eyes of the public.62  In other words, the credibility of a scientist and the credibility of their work are not neatly separable entities within the public mind. If you are able to undermine one, you will undermine the other without having to wade into the actual science.

Anthony Fauci is one of the cleanest contemporary examples of how scientific opinions that interfere with policy preferences can become a personalized series of attacks in the United States. During the 2020 COVID-19 pandemic, Fauci quickly rose to prominence in the public eye for his role as the director of the National Institute of Allergy and Infectious Diseases,63  quickly becoming one of the most prominent—and infamous—scientific officials in the history of the country.

In January 2025, President Trump terminated the federal security protection Fauci had continued to receive after leaving government due to ongoing threats against him, reinvigorating streams of attack against him.64  Several months later, the White House replaced the federal government’s COVID-19 information website with a page presenting the administration’s account of the pandemic. Rather than simply presenting an alternative interpretation of particular policies or pieces of evidence, the site explicitly centered Fauci—alongside former President Joe Biden and the World Health Organization—as a target of criticism in its account of failures surrounding the pandemic response.65

Scientists are not immune from criticism, and findings should be challenged as a fundamental principle of the scientific process. However, by attacking the moral character of a figure like Fauci, Trump and similar-minded politicians can avoid wading into the overwhelming evidence that COVID-19 precautions and vaccines saved lives66—evidence which runs contrary to the policy preferences of the Trump administration.67 

Once scientists have been thoroughly questioned, a given administration sets itself up nicely to argue in favor of replacing previous scientists with political ideologues—which we cover in the following section. 

Replacing Experts with Loyalists

Replacing experts with loyalists is often the flip side of attacking the scientists. By pulling into question the legitimacy of past scientists using moral attacks, it becomes easier to provide an argument for their replacement—often one who is more ideologically-aligned with the politicians of the present. For example, by the 1930s, Lysenko (detailed in the previous section) ended up rising from relatively obscure beginnings to becoming one of the most powerful scientific figures in biology during Soviet Russia, because his ideas were deeply attractive to Soviet political leadership since they fit neatly into the broader ideological belief that nature, like society, can be rapidly transformed under the optimum conditions. His political support allowed him to accumulate increasingly powerful positions within Soviet science, eventually becoming the director of the Institute of Genetics of the Soviet Academy of Sciences (which is now, ironically, called the Vavilov Institute of General Genetics).68  At the infamous 1948 meeting of the Lenin All-Union Academy of Agricultural Sciences, Lysenko presented his views as the scientifically and ideologically correct theory of heredity, announcing the Communist Party approved of his report. Classical genetics was subsequently pushed out of Soviet research and education.69 

Expertise is produced through rigorous training, accumulated knowledge, empirical scrutiny, and demonstrated competence within a certain field. Scientific authority should therefore be, at least in substantial part, epistemic: the people given the greatest influence over scientific questions should have relevant expertise, and should remain accountable to the standards of evidence within their particular discipline. The relevant question is therefore not whether a given scientist’s positions agree with scientific consensus, but rather, whether their authority is meaningfully derived from relevant expertise and the strength of evidence they put forth—or, alternatively, if their authority is merely a byproduct of their position proving sufficiently useful for those in power to appoint them. 

When ideological compatibility begins to substitute for disciplinary expertise as a criterion for scientific authority, the consequences extend beyond which scientific positions are elevated. Expertise is also part of the process through which evidence is evaluated: determining which information is relevant, assessing the strength of competing findings, recognizing methodological limitations, and deciding what conclusions the available evidence can reasonably support. Replacing expertise with political compatibility therefore risks weakening not only which conclusions are treated as authoritative, but the rigor of the process used to arrive at those conclusions in the first place.

One of the most prominent examples of this process under the second Trump administration concerns the CDC Advisory Committee on Immunization Practices.70  On June 9, 2025, Trump-appointed Secretary of Health and Human Services Robert F. Kennedy Jr. removed all 17 sitting members of the committee at once.71  Two days later, Kennedy announced eight hand-picked replacements. Among these was Robert Malone, who had opposed the use of mRNA vaccine technology, alongside other appointees who questioned aspects of existing vaccine policy.72  STAT additionally reported that several appointees had little to no vaccine expertise,73 and a subsequent Science analysis found that the new members had significantly less vaccine-science expertise than the panel they replaced.74 

The original ousted panelists additionally publicly warned that Kennedy’s overhaul erodes the credibility of vaccine policy and public trust in immunization decisions.75  Naturally, however, the HHS described the decision to completely revamp the panelists in exactly the opposite terms— the department said the overhaul was intended to eliminate conflicts of interest, restore public trust, and ensure that “unbiased science” guided vaccine recommendations. Interestingly, as a part of a broader overhaul, HHS now identifies the principles of “Gold Standard Science” as an organizing principle, including what it calls “constructive skepticism of findings and assumptions.”76  While skepticism is an inherent part of the scientific process, it can easily be weaponized to manufacture uncertainty on well-established scientific knowledge, which we turn to next.

Manufactured Uncertainty On Established Knowledge

In the 1950s, a rapidly growing body of scientific evidence was beginning to show irrefutable evidence that cigarette smoking was linked to lung cancer. This was an obvious incoming threat to the U.S. tobacco industry, with fears of incoming public health interventions against the use of cigarettes motivating the industry to pursue tactics to stifle public fears on smoking. To do so, the industry recruited a band of scientists who were willing to challenge the findings that linked cigarette smoking to disease, while simultaneously drawing attention to the limitations and range of unanswered questions that still remained after considering the existing evidence. In doing so, the tobacco industry did not have to prove that cigarettes were safe—an impossible task—but rather, they simply had to sow doubt and convince smokers that evidence showing cigarettes could cause cancer was not yet settled.77

It was surprisingly effective. The work by the U.S. tobacco industry to sow doubt helped preserve uncertainty around cigarette smokers, even after the scientific case was too strong to overlook. In 1966, a Public Health Service survey found that over 60% of smokers were still under the impression that the relationship between cigarettes and cancer was “not yet proved”, and shockingly, more than half reported that they would not fully be convinced that cigarette smoking was actually harmful until the tobacco industry itself admitted it.78  Notably, these opinions were recorded after the U.S. Surgeon General issued a landmark 1964 report formally declaring that cigarette smoking causes lung cancer and other serious diseases.79

The work by the U.S. tobacco industry demonstrates our fourth pillar: an established scientific conclusion does not need to be wholly disproven to lose its influence—it may be enough to convince the public that the science is still uncertain. 

As long as the scientific method continues, manufacturing uncertainty about well-established findings will be a possible line of attack for bad actors. The nature of science is that it rarely speaks in absolutes—findings are derived from approximations with margins of error, and even robustly demonstrated findings are often framed as theories, not conclusive evidence of how the world operates. It is essential in the scientific process that scientists are capable of acknowledging what is not yet known, can identify possible limitations of their work, and be able to update theories when new information is discovered. However, the uncertainty that exists within the scientific enterprise can also be corruptedly magnified in a way to make seemingly well-established conclusions appear far less certain than they realistically are—a process Dr. David Michaels (an epidemiologist at George Washington University) calls “Manufactured uncertainty”.80

One particularly stark example of this line of attack during the second Trump administration is in regards to the 2009 Endangerment Finding.81  In 2009, the EPA found that six greenhouse gases in particular threatened public health, as well as the welfare of both current and future generations of Americans. This finding became the backbone of many federal regulations on greenhouse gas emissions. However, In March 2025, Trump-appointed EPA Administrator Lee Zeldin publicly announced that the EPA would be reconsidering the Endangerment Finding, drawing explicit attention to the “significant uncertainties” he claimed were present in the science behind the original finding. To address this, the EPA commissioned the Department of Energy to assess the supporting climate science. Lo and behold, the DOE challenged the scientific basis underlying the 2009 Endangerment Finding, a challenge which was then used to support the EPA eventually rescinding the 2009 finding in February 2026.82

The science did not change—if anything, subsequent climate research following the 2009 finding has documented increasingly clear harms to human health and welfare. Manufacturing uncertainty on established knowledge is about amplifying desired findings, while suppressing unwanted contradictory information—it is a game of optics, not pursuit of ultimate scientific truth. This is best displayed by none other than Trump saying in 2024—prior to the DOE’s investigation of the GHG findings—he would pursue regulatory rollbacks for his polluter donors at events hosted by major industry figures.83 

Like the administration’s slogan of efficiency and eliminating “fraud, waste, and abuse,” they have also co-opted the language of accountability to inject uncertainty in science. In May 2025 the White House issued Executive Order 14303, “Restoring Gold Standard Science,” directing agencies to rewrite the rules that govern how scientific information is produced and used so they match the order’s nine tenets. The language—reproducibility, transparency, uncertainty, peer review—almost sounds reasonable until it is read in the context of the power that sits with political appointees, who can now certify what meets the standard and what can be dismissed as not rigorous enough. That is how the “gold standard science” has  become a tool to delegitimize work the administration does not want. Prior to the release of EO 14303, the NIH rescinded its final Scientific Integrity Policy,84  which had only been finalized just under six months prior. It is then unsurprising that, following EO 14303, the NIH released an agency-wide plan to instead implement the nine tenets of “Gold Standard Science” across its institutes.85 

Thus far, we have covered four major pillars. The first two are primarily aimed at past work (Removing Access to Information and Attacking the Scientists), while the second two largely concern present policy (Replacing Experts with Loyalists and Manufactured Uncertainty on Established Knowledge). For our fifth and final pillar, we turn to the future. Political control over science can extend well into the future, shaping which questions are permitted to ask, and as a result, determine who is allowed treatment, whose future we will protect, what public health concerns will be addressed, and what emerging problems we have the opportunity to understand before they become public crises. 

Controlling What Questions Can Be Asked

In 2018, the Hungarian government—under Viktor Orbán—deleted gender-studies graduate programs from the list of possible degrees offered by their universities.86  Members of Orbán’s government had a broad objection to the field, arguing that gender studies represented an ideology more than a legitimate academic discipline. The government in doing so weaponized its control over university accreditation to shape what academic disciplines are viewed as legitimate. The impact of this is not a deletion of access to gender-related studies, but rather, a discontinuation of researchers who specialize in gender-studies entering the workforce. A government may not need to directly suppress inconvenient or unwanted scientific findings that may threaten its policy preferences if it can halt the question from being asked in the first place.

This constitutes our fifth and final pillar: controlling what questions can be asked. The most effective way to suppress an inconvenient scientific finding may be to prevent the evidence from being produced in the first place. Scientific knowledge depends on more than the freedom to publish: researchers need funding, institutional support, access to data, labs, trainees, and years in which to pursue a question. Governments that control many of these resources therefore possess the influence to  determine which questions may be funded or which fields are institutionally recognized at all. Political power can shape not merely how science is communicated but what knowledge or lines of inquiry should be pursued.

The pattern in the United States under the second Trump administration bears similarity to Hungary. In early 2025, the National Institutes of Health terminated more than 2,000 grants in just five months, with a large swath of the affected research and projects involving health equity and sexual and gender minority populations.87  A separate study found 64 canceled previously NIH-funded research grants, with specific subfields included examining HIV, cardiovascular disease, cancer, and other life-or-death illnesses in transgender populations.88  Of particular importance here is the language used to justify the termination notices, with projects involving work including diversity, equity and inclusion, gender identity, vaccine hesitancy, and other subjects terminated for being “no longer aligned with agency priorities.”

Again, it is not abnormal for federal governments to set research priorities at large. The concern for scientific independence arises, however, when the acceptability of the question itself is contingent on its political compatibility, rather than its scientific merit. This will additionally impact what research grants are proposed for years to come, as scientists preparing for the next round of grant applications can see which studies are getting terminated, what terms are attracting scrutiny, and what areas will not be supported altogether. Over time, these decisions influence the future of scientific literature by way of what information is being collected today.