The assault on universities and on the institutions of public reason that has unfolded between January 2025 and the spring of 2026 is not a contingent excess of an unusually disorderly administration but a structural feature of competitive authoritarianism in its anglophone instantiation. The assault is, in the precise sense of Levitsky, Way and Ziblatt (Levitsky and Way, 2010, 2025, 2025), an attack on the soft guardrails of democracy: the unwritten norms of mutual toleration and institutional forbearance whose erosion is the proximate cause of democratic death-by-stages.
Two corollaries follow, and they have to be held together. The first is that the United States in 2025–2026 is already an authoritarian polity in the sociological sense in which that term has a settled comparative referent. The second is that this authoritarianism is non-consolidated — that elections are still contested, the judiciary still rules against the executive often enough to matter, federalism still generates rival centres of authority, and the November 2025 off-year elections demonstrated that the opposition can win (Levitsky and Way, 2025). The analytic stance that follows is therefore neither alarmism nor reassurance, but the double vision that competitive authoritarianism specifically demands: the threat is real, and the contest remains open.
2.0.2 The assault on the university: a sequence
The chapter reconstructs the 2025–early-2026 sequence in close detail because the speed of events has tended to outpace academic analysis, and because the cumulative pattern is more revealing than any single episode.
Q1 2025 — opening salvos. Title VI investigations against Columbia (March), Penn, Brown and Cornell open in rapid succession, joined within days by an additional fifty-one investigations targeting race-based scholarship and DEI programmes (Spitalniak, 2026). The Department of Education’s anti-DEI guidance of February 2025 — subsequently struck down in federal court in August — provides the rhetorical cover. Demand letters carry extraordinary scope: external receivership of academic departments, prohibition of mask-wearing in protest, viewpoint-diversity quotas in faculty hiring. By the end of the first quarter, the administration has opened or cited more than 150 investigations into US colleges and universities (Spitalniak, 2026).
Q2 2025 — escalation. Announcement of the Catch and Revoke programme, deploying machine-learning systems against international students’ social media (Bulletin of the Atomic Scientists, 2025). Threatened cap on indirect research-cost reimbursement at fifteen per cent — pursued in nearly identical form by NIH, NSF, the Department of Energy and the Department of Defense, and subsequently blocked by federal courts in all four cases (Spitalniak, 2026). The administration’s fiscal-year 2026 budget request proposes a twenty-one per cent cut to federal scientific research, with NSF reduced by more than half and NIH by nearly forty per cent (Spitalniak, 2026); congressional appropriators reject the proposal and pass bipartisan alternatives. Harvard sues the administration; lower courts begin to rule against its procedural unlawfulness, including the attempt to revoke the university’s authorisation to enrol international students. In June, the President addresses uniformed soldiers at Fort Bragg and goads them to jeer at elected Democratic officials (Levitsky and Way, 2025) — a small breach in the firewall between the armed forces and partisan politics whose significance is that it was attempted at all.
Q3 2025 — institutional sorting and political shock. Columbia capitulates under sustained financial duress and accepts a sweeping settlement: a payment of $221 million, the largest of any agreement to date, together with extensive policy concessions including reporting of admissions data to the federal government, mandatory ‘socialisation’ training for students and an independent monitor overseeing compliance (Spitalniak, 2026). Other private universities — MIT, Brown, USC, Penn, UVA, Dartmouth — initially reject the proposed Compact for Academic Excellence in October. In September, the right-wing commentator Charlie Kirk is shot and killed; in the days that follow, dozens of teachers, university faculty and journalists are suspended or dismissed for social-media commentary, including, as Levitsky, Way and Ziblatt note, those whose only offence was to criticise Kirk’s published positions (Levitsky and Way, 2025). Later that month, the President instructs senior military commanders to prepare for deployment in US cities against an ‘enemy from within’ — a register, the same authors note, of the language used by the South American military juntas of the 1970s.
Genealogy of the ‘enemy from within’:
The rhetorical infrastructure for this register was laid in June 2020, when the administration characterised protesters as ‘terrorists’ and prepared the use of active-duty troops in US cities under the Insurrection Act (Hill, 2020). The structural homology with the doctrina del enemigo interno — the internal-enemy doctrine through which Southern Cone juntas in the 1970s justified the criminalisation of domestic dissent and the granting of broad impunity to security forces — became explicit five years later, when the federal deployment in Los Angeles in June 2025 was identified by human-rights observers with regional expertise as reproducing precisely that doctrine (Washington Office on Latin America, 2025). The September 2025 escalation, with its ‘enemy from within’ formulation, is therefore neither rhetorically novel nor operationally improvised: it is the doctrinal completion of a sequence whose first articulation is half a decade old.
Q4 2025 — settlements, litigation, and electoral correction. By year’s end, six universities have brokered public settlements with the administration: Brown, Columbia, Cornell, Northwestern (at $75 million in late November), Penn and the University of Virginia (Spitalniak, 2026). Most paid hundreds of millions in combined settlement and reform costs to recover frozen federal research funding; UVA, whose funding remained intact, traded investigatory closure for a written commitment to terminate DEI through 2028. Harvard, Princeton and Cornell continue litigating the broader pattern, and a federal judge in Massachusetts strikes down the administration’s $2.2 billion funding freeze against Harvard (Spitalniak, 2026). In the November off-year elections, the Democratic Party wins decisively, demonstrating that elections in the United States remain — in the technical sense the competitive-authoritarianism literature requires — competitive (Levitsky and Way, 2025).
Q1–Q2 2026 — paramilitarisation and corporate fracture. In January 2026, the conversion of Immigration and Customs Enforcement into what observers and former officials have called a ‘poorly regulated paramilitary force’ produces civil unrest in Minnesota and triggers a letter from roughly 2,500 Silicon Valley employees demanding their employers sever contracts with the agency (elDiario.es, 2026). In April, the federal appeals court permanently blocks NIH’s indirect-cost cap (Spitalniak, 2026). In the same window, the Compact for Academic Excellence is variously revived, modified and contested across institutional negotiations. The full fiscal-year-2026 budget remains unresolved past the late-January stopgap.
The chapter’s analytical claim about this sequence: the assault is institutionally targeted, financially leveraged and rhetorically masked. It is targeted because universities are not collateral damage but a designated front. It is financially leveraged because the operative coercion is the freezing of already-appropriated federal research funding, not the prohibition of any specific activity — a tactic precisely chosen to compel settlement before legal review can run its course. And it is rhetorically masked because the language of antisemitism, viewpoint diversity and DEI is instrumentalised to displace legal-procedural constraints: the substantive assault is procedural and financial; the public justification is moral and cultural. The asymmetry between the operative and the justificatory registers is the precise diagnostic of the regime type.
2.0.3 The transnational pattern: Applebaum’s Autocracy, Inc. and its discontents
The analysis then steps outward. Applebaum’s Autocracy, Inc. (Applebaum, 2024) argues that contemporary autocrats constitute not a bloc but a network — bound not by ideology but by mutual interest in the suppression of independent civil society, the protection of kleptocratic wealth and the propagation of disinformation. Three features of her account matter here.
The first is the post-ideological character of the network. Russia, China, Iran, Venezuela, Hungary, El Salvador and others share neither communism nor any other unifying doctrine; what they share is a method. The American case fits this template uneasily, because it carries the additional weight of an ideological tradition — libertarian-conservative, evangelical-nationalist — that none of the other regimes possess in comparable form. But the method is recognisable: capture of the regulatory state, weaponisation of tax and legal authority against opponents, financial leverage over civil society, and a pliable supportive media ecosystem.
The second is the structural complicity of Western financial and professional infrastructure. The City of London, Wall Street, Sussex real-estate agents, Sioux Falls bankers — Applebaum’s catalogue of enabling infrastructures is precise, and it is here that Slobodian’s work on the cross-fertilisation of US conservative intellectual networks with international authoritarian capital adds analytical depth (Slobodian, 2025). The argument here — and developed further in §2.5 — is that the United States has, between 2024 and 2026, moved from being one of the principal targets of Autocracy, Inc. to being a partial participant in it: an inversion that is constitutively related to the central claim about the assault on universities, because universities are precisely the institutional sites at which independent expertise about kleptocratic finance, climate science and election integrity is produced.
The third is the role of disinformation as a mode of governance. Modern autocratic propaganda does not seek to persuade but to disorient — to overwhelm citizens with mutually inconsistent claims until they retreat into private cynicism (Applebaum, 2024; Harsin, 2024). The American variant has a particular texture, captured well by Trujillo and Trujillo’s finding that distrust of experts is morally rather than competence-driven (Trujillo and Trujillo, 2026): it is not enough that experts be wrong; they must be bad. This is the rhetorical condition that allows a freeze of $2.2 billion in research funding to be presented as a moral act.
A complication that Applebaum’s frame cannot, on its own, hold, must be registered here. The American technology sector is neither a unitary participant in the authoritarian project nor a coherent opposition to it; it is fractured along precisely the lines the project requires for its operation. The fracture has a paradoxical shape that bears emphasising: the corporations whose competitive advantage rests on cognitive infrastructure — on the recruitment of the most highly trained scientific and engineering personnel available, the great majority of whom were formed in the very research universities now under coercive financial pressure — have, at the level of executive decision, sided with the administration that is dismantling the conditions of their own technical reproduction. The clearest illustration is the early-2026 confrontation between Anthropic, Palantir and internal dissenters at Google (elDiario.es, 2026). Anthropic refused contracts with the Pentagon for what it identified as two red lines — autonomous lethal weapons and mass surveillance — and was placed by the administration on a ‘supply-chain risk’ blacklist; the punishment failed when a new Anthropic model proved strategically indispensable to US cyber-security, forcing the White House to reverse course (elDiario.es, 2026). Palantir, by contrast, published a twenty-two-point manifesto demanding the dissolution of any ethical limit on military AI, framing the contest in explicitly civilisational terms (Down and Booth, 2026; Karp and Zamiska, 2025). More than 560 Google employees, including some twenty vice-presidents, signed a letter to Sundar Pichai demanding the company decline a new Pentagon contract; the contract was signed regardless (elDiario.es, 2026).
The argument is not, to be clear, that Silicon Valley redeems itself: the contracts were signed, ICE was paramilitarised, and the surveillance infrastructure remains in place. The argument is that the alliance between corporate AI leadership and an administration committed to dismantling the institutions of independent research is self-undermining: it sacrifices, for short-term contractual advantage and protection from regulatory hostility, the long-term institutional conditions on which the sector’s technical superiority depends. That this self-undermining quality is internally legible to a meaningful share of the technical workforce — including senior managers — is itself a political resource of the kind the competitive-authoritarianism literature treats as a constitutive feature of the regime type (Levitsky and Way, 2010).
The self-undermining quality of this alliance is structural, not incidental, and bears specifying. The competitive advantage of US technology firms in the global market for advanced AI rests on three institutional preconditions, none of which is a market output: the freedom of scientific inquiry, the unrestricted flow of ideas and personnel across institutional and national borders, and a regulatory environment of civil-rights guarantees that makes the United States more attractive to top-tier researchers than its principal geopolitical competitors. The recruitment advantage that US firms enjoy over Chinese and Russian counterparts — high salaries, intellectually demanding environments, freedom to publish and to associate — is not incidental to that ecosystem; it is that ecosystem. An administration that defunds federal research, revokes the visas of international students for political speech, weaponises Title VI investigations against the leading research universities, and deploys ICE against immigrant academics and engineers is, by the same operations, eroding the institutional substrate on which the sector’s global preeminence is built (Spitalniak, 2026).
What makes the Palantir position particularly revealing is that it articulates the contradiction as a programme rather than a silence. The Technological Republic (Karp and Zamiska, 2025), whose twenty-two-point manifesto (Down and Booth, 2026) crystallises the corporate stance, frames elite research universities as a source of the problem rather than the source of the technical capital on which the sector depends. They cultivate, on this account, a generation of ‘technological agnostics’ insufficiently committed to the national project; their hold on the formation of the engineering elite is a liability rather than an asset. The argument has internal consistency; it also has the demerit of mistaking the conditions of one’s own existence for the obstacles to it.
The conditions in question are not abstract. The elite universities Karp and Zamiska arraign are the institutions that trained Karp himself, that supply Palantir its engineers, and whose graduate research traditions in machine learning, optimisation and applied statistics are the precondition of the company’s products. To call for the political subordination of those institutions while continuing to depend on their output is not a coherent strategy; it is a wager that the institutional capital accumulated over a century will outlast the political project designed to dismantle it.
2.0.4 Why universities? An institutional-political answer
A closing analytical section asks why universities, specifically, have been the front. The standard reasons converge with two that the literature has been slower to articulate.
Universities are concentrations of organised counter-power. They are wealthy, durable, and host professional norms — peer review, academic freedom, tenure — that are constitutively resistant to political capture. Of all the soft-guardrail institutions Levitsky and Ziblatt identify, universities have the longest institutional memory and the deepest reserves of legitimacy. The United States hosts more than 1,700 private institutions of higher education (Levitsky and Way, 2025); their aggregate financial, organisational and reputational capital exceeds anything available to oppositions in Hungary, India or Turkey, let alone Venezuela or Russia. They are, in this sense, an unusually well-armed civic defence.
They are sites of the production of legitimating expertise. Climate science, public-health expertise, historical scholarship, legal scholarship: all originate in universities and constrain the rhetorical space available to authoritarian governance. The agenda articulated by McKee and colleagues for public-health diplomacy in the populist age makes the point with unusual clarity: when populist governance requires the suppression of inconvenient evidence, the institutional sources of that evidence must be either captured, defunded or de-legitimated (McKee et al., 2025). The proposal to cut federal scientific research by twenty-one per cent (Spitalniak, 2026) is not a budgetary measure independent of the political project; it is the budgetary expression of the political project.
They are emotionally salient targets. The cultural symbolism of ‘elite universities’ lends itself to populist mobilisation in a way that, say, central banks or constitutional courts do not. Their visibility is a political asset for those who attack them. The trope of the intellectually corrupt, financially over-endowed, and ideologically homogeneous university operates at the level of public imagination with such fluency that even the recoil from the specific tactics — funding freezes, ideological compacts, visa revocations — has not unmade it.
To these familiar reasons, two further considerations must be added. The first is the epistemic dimension recently developed by Heumann’s analysis of rural consciousness in Switzerland, which is generalisable to the comparative literature (Heumann, 2026). Heumann argues that what the literature has read as an affective divide between rural and urban populations is more accurately read as an epistemic divide: rival frameworks of knowledge — proximity, hands-on practical work, common sense — that disqualify expert, abstract, academic knowledge rather than merely distrusting it. This is more fine-grained than the conventional rural-resentment account, and it has a sharp implication for the argument here: the rhetorical attack on universities lands on a population already predisposed to read expert knowledge as a category mistake about where authoritative knowledge comes from. Merkley’s earlier work on anti-intellectualism and motivated resistance to expert consensus (Merkley, 2020) anticipates the same finding from a different angle. Berg, working on anti-COVID protest, adds a further refinement: such protesters do not, on closer inspection, reject science but invoke it, deploying the rhetoric of evidence and citation against the institutional carriers of expert authority (Berg, 2025). This is not anti-science as the simple mirror of science; it is anti-science as a re-coding of the same rhetorical resources.
The second consideration sharpens the picture by complicating the defensive register the chapter has thus far adopted. Universities are not innocent. The internal anti-intellectualism that managerial discourse has installed in the academy over the past four decades is structural, not merely rhetorical. Three strands of analysis converge on the same diagnosis from different angles. Ginsberg’s account (Ginsberg, 2011) tracks the demographic mutation of the modern university into what he calls the ‘all-administrative’ institution: an organisational form in which the growth of administrative personnel has exceeded that of faculty by a wide margin and in which faculty governance has been progressively displaced. Newfield’s (Newfield, 2016) political-economic account documents the parallel privatisation of the US public university — declining state appropriations absorbed through tuition increases, dependence on private and corporate funding, adoption of the rhetorical and managerial apparatus of the for-profit firm — as a four-decade self-inflicted wound on the institutional segment that historically anchored American mass higher education. Diaz’s glossary (Diaz, 2021), drawing on Adorno’s Halbbildung, catalogues the linguistic surface of the same process: the rhetoric of ‘excellence’, ‘leadership’, ‘efficiency’, ‘human capital’ and ‘impact’ through which the academy has internalised the vocabulary of the firm.
The point is not to flatten the asymmetry between an external coercive assault and internal deformations — the asymmetry is real, and the chapter does not equivocate on it — but to note that the assault has succeeded in part because the institutional self-confidence required to withstand it had already been hollowed out by four decades of managerial reform. A further consequence worth flagging: the 2025 assault has been politically possible on the private segment of US higher education in part because the public segment had already been weakened by the long privatisation that Newfield documents (Newfield, 2016), leaving the federal government with a smaller field of public resistance than the mythology of American higher education suggests.
A related caveat is owed to the rhetorical instrumentalisation of ‘viewpoint diversity’. McDonald has argued, persuasively, that the actually existing American research university already exhibits substantial viewpoint diversity at the level that matters — the disciplinary level, where economists, comparative literature scholars, philosophers of mind and historians of medicine do not in fact share a worldview, much less a hegemonic ideology (McDonald, 2026). Federal and state programmes purporting to introduce viewpoint diversity into universities are therefore not addressed to a real condition; they address a phantom and impose political constraints under cover of doing so. The point is analytically important because it specifies what the assault is not: it is not a corrective response to an ideological capture of the academy, because that capture has not, at the disciplinary level, occurred.
The chapter’s analytical pay-off, then, is this: the present assault on universities is not aberrant or anti-intellectual in passing — it is the precise operational expression of competitive authoritarianism’s logic in contexts where universities are still institutionally autonomous enough to matter, addressed to a population whose epistemic dispositions have been pre-formed to receive the rhetoric, and finding its targets already weakened by four decades of internal managerialism. None of these factors alone would be sufficient. Together they constitute the political condition that the rest of the book interrogates.