French Counter-Disinformation Practitioners See a Supply-Side Threat, but Say Institutions Are Failing to Keep Up
PARIS — The officials, military officers, regulators, and journalists responsible for fighting disinformation in France share a strikingly consistent diagnosis of the problem: the threat comes from the top, not the bottom. Platforms, algorithms, and organized producers are seen as the primary drivers of false content, while ordinary citizens are viewed less as culprits than as vulnerable targets. Yet this convergence in diagnosis does not translate into effective action. In a new survey of 106 French counter-disinformation practitioners, published by researchers Grégoire Darcy of the Institut Jean Nicod and Sacha Altay of the University of Zurich, three-quarters of respondents said that governments proclaim the fight against disinformation a priority without providing the resources, coordination, and clear mandates needed to win it. The study, which draws on senior figures from the offices of the President and Prime Minister, the foreign affairs and health ministries, military and intelligence services, independent regulators such as ARCOM and ARCEP, and specialized journalists, offers one of the first systematic portraits of the people who shape democratic responses to information manipulation. Unlike previous surveys of scientists and academics, it focuses on the practitioners who act on research, often under conditions of fragmented authority, ambiguous legal mandates, and intense political sensitivity. The findings, based on a 92-item questionnaire fielded in French between September 2025 and February 2026, paint a picture of a field that is acutely aware of the scale of the threat but structurally unable to meet it.
The practitioners surveyed do not merely consider disinformation a serious issue; they rank it as a first-order societal threat on par with democratic decline and climate change. Ninety-six percent rate disinformation as an important or very important societal problem, with most placing it at roughly the same level as democratic decline, climate change, and armed conflict. Large majorities attribute broad macro-political consequences to it: 90 percent view foreign campaigns as a major threat to democracy, 87 percent believe disinformation sustains Russian support for the war in Ukraine, 81 percent say it fuels vaccine hesitancy, 80 percent believe it enables voter manipulation, and 78 percent say it erodes trust in the state. Perhaps most strikingly, 83 percent agree that a disinformation campaign can decisively change an election result, although agreement drops to 71 percent on the specific case of Donald Trump’s 2016 victory, suggesting that practitioners see the threat as greater in the abstract than in concrete instances. Their diagnosis is fundamentally supply-side. Nine in ten practitioners rate social media platform owners as a major threat to the information environment, and nearly three in ten name them the single greatest threat. Foreign governments rank second at 69 percent, followed by private media at 63 percent and domestic politicians at 57 percent. Ordinary citizens come last by a wide margin, with only 18 percent viewing them as major drivers. The same logic shapes their beliefs about spread and belief: 85 percent say platform algorithms strongly influence political opinions, 84 percent point to algorithmic promotion as a key reason people believe disinformation, and 76 percent say producers and platforms distribute misleading content on a massive scale. By contrast, the idea that the public actively seeks out such content divides the sample almost exactly in half, with 39 percent agreeing and 43 percent disagreeing. When asked why people believe disinformation, practitioners emphasize confirmation bias and selective exposure at 90 percent, followed by emotional reactions, repetition, and identity-based motivations. Notably, institutional distrust occupies a relatively modest place in their thinking, even though academic research often treats eroded trust in institutions as a central condition of vulnerability to misinformation. Compared with a 2022 survey of academic scientists, practitioners agree with researchers on well-established psychological mechanisms such as confirmation bias and repetition, but diverge sharply on contested questions: they rate the electoral impact of misinformation far higher than scientists do, and give significantly less weight to institutional distrust, which scientists treat as central. The authors describe this as a selective diagnosis, one that overweights supply-side infrastructure and individual credulity while underweighting the social and institutional conditions that make misleading claims persuasive in the first place.
This shared diagnosis does not, however, translate into a shared capacity to act. The survey asked practitioners to rate the obstacles they face, and the top answers reveal a field blocked less by doubt about the problem than by the structure of the response itself. The single most important obstacle, rated very or extremely important by 76 percent of respondents, is a gap between political rhetoric and concrete action. The complaint comes from inside the state: officials, officers, and regulators report that counter-disinformation is proclaimed a priority without the budgets, staff, coordination, and indicators an effective response requires. The second and third most cited obstacles are reactive policies that lag behind new tactics, at 71 percent, and the influence strategies of large digital companies, at 70 percent. A cluster of structural constraints follows, each cited by a majority of respondents: insufficient budgets and staff, administrative slowness, lack of operational tools and shared indicators, unclear roles and coordination with no lead actor, and difficulty attracting and retaining specialists. Perhaps most telling is the role of censorship accusations. Sixty-three percent of practitioners rank the fear of being accused of censorship among their most serious obstacles, yet only 21 percent say they encounter actual ethical dilemmas often in their daily work, while 62 percent encounter them rarely or never. The constraint is therefore largely anticipatory: what inhibits practitioners is not moral conflict they have experienced, but the accusation they expect to face if they act. This fear, the authors note, is deeply embedded in the institutional logic of the field. Because disinformation is politically sensitive and because responses to it can look like speech restrictions, practitioners often hesitate to intervene, even when they believe intervention is justified. The result is a system that is simultaneously convinced of the severity of the threat and paralyzed by the legitimacy costs of acting against it. The authors also found that perceived obstacles vary by educational background, with practitioners trained in computer and data sciences rating obstacles as less severe overall than those from social sciences, humanities, and communication backgrounds. A dedicated comparison of public-sector and private respondents found no detectable difference in their substantive diagnosis, although the precision of that comparison was limited.
The relationship between practitioners and scientific research adds another layer of constraint. The survey found that practitioners trust science but rarely engage with it directly. Seventy-one percent say they trust peer-reviewed research to guide their decisions, but only about a third report keeping up with recent academic work. Nearly all respondents, 88 percent, read about disinformation primarily in French, a language in which far less of the scientific literature is available, and their most frequent sources are the press and gray literature, with scientific articles playing a secondary role. The barriers to deeper engagement are overwhelmingly organizational rather than intellectual. Lack of time, limited access to publications, and weak institutional incentives all rank above low perceived applicability, the language barrier, or methodological difficulty in practitioners’ self-reports. The authors caution that these self-reports may understate the role of methodological difficulty and scientific English, since admitting difficulty with methods is more face-threatening than citing lack of time. But the broader pattern is clear: research is trusted in principle but consumed through intermediaries, which means practice is largely dependent on secondhand interpretations of the evidence base. This gap between trust and engagement has consequences. The authors argue that practitioners’ selective diagnosis, particularly their overestimation of electoral effects and their underestimation of institutional distrust, may reflect incomplete uptake of the evidence. On elections, the best available research finds little sign that misinformation decides outcomes. On trust, comparative research treats institutional trust as central to vulnerability, and low institutional trust is widely recognized as making people more receptive to misinformation, even by researchers who disagree about misinformation’s broader societal effects. The authors also consider a second reading: that the available evidence, which is largely Anglophone, platform-centric, and focused on average population effects, may say less about the targeted foreign operations that many French practitioners confront. A third reading ties the other two together. Institutions tend to diagnose problems in terms of what they can measure. Disinformation’s effects are hard to measure, and prevention that works leaves no trace, since success means the expected harm never happens. So institutions count what they can see instead: the attempts they detect, mostly online. That focus on the digital, upstream side steers action toward reactive policies whose results can be shown, and acting at that level makes the technology-centered diagnosis look right, confirming itself. This measurement loop, well documented in public administration, is unusually strong in counter-disinformation because its ultimate goal, an information environment citizens can trust, is among the hardest outcomes to measure, while its operational objectives, campaigns detected and content removed, are easier to count.
The future outlook among practitioners is grim. Nearly all respondents expect disinformation to become a greater threat over the next three to five years. Ninety-six percent anticipate increases in foreign interference, 93 percent expect more false or manipulated content in ordinary users’ feeds, and 83 percent expect declining trust in media, science, and institutions. Yet on the response side, confidence is strikingly low. Six in ten practitioners think state capacity will increase, but only one in eight, 13 percent, believe that future capacity will be sufficient to contain the threat, while more than two in three judge that outcome unlikely. This is not a portrait of a field that sees a clear path forward. The authors note that practitioners see no trajectory that closes the gap between the growing threat and the institutional response under current arrangements. Layered on top of this pessimism is a diversity of mandates and objectives. Practitioners face different threats, use different tools, and define success differently. An educator measures success by accurate beliefs, a regulator by compliance, and a military officer by attribution of hostile operations. The survey found that educational background is the most consistent source of variation in how practitioners perceive obstacles to action, and that fear of censorship accusations varies by role, with the highest levels among normative, regulatory, and public-protection roles and the lowest levels among cybersecurity practitioners. This diversity, the authors argue, means there is no single standard for fighting disinformation, and the evidence base is just as fragmented. Psychology and communication studies focus on corrections, literacy, and inoculation; security, intelligence, and computer science focus on attribution and deterrence; and some problems, like evaluating institutional responses, fall to no field at all.
The study’s implications point toward institutional reform, better knowledge translation, and a more honest reckoning with the limits of current evidence. For institutions, the authors recommend that coordination should accommodate the diversity of roles rather than flatten it, and that a limited set of common indicators, covering exposure, incident response, coordination quality, and effects, embedded in clear mandates and auditable safeguards, could offer a way out of the measurement loop. For researchers, the findings suggest that the field needs to translate existing research into practitioners’ working language, through rapid briefs, short syntheses, and permanent brokerage roles that connect studies to training, doctrine, and operational decisions. The reverse flow also matters: practitioners’ operational questions should be carried back into research agendas, and the evidence base should be extended beyond the Anglophone, platform-centric, average-effect studies that dominate current literature. The authors offer one speculative arrangement: a standing observatory of the information environment that would combine measurement and knowledge translation, building the indicators current instruments miss and deploying the knowledge translation mechanisms that the evidence-use literature has tested. They also stress that transparent safeguards are essential to make legitimate intervention defensible in the face of censorship accusations. The French setting, they note, is both specific and instructive. As a European Union and NATO member, France is covered by the Digital Services Act, into which the Code of Practice on Disinformation was integrated in 2025, and it has built one of Europe’s most institutionalized counter-disinformation apparatuses, including VIGINUM, the government service against foreign digital interference, and a national strategy adopted in 2026. This makes France both a leading case and an increasingly referenced model for partners. At the same time, France concentrates features common to many democracies: fragmented responsibility, strong security salience, and uneven access to an overwhelmingly Anglophone literature. The authors are careful to note the study’s limitations: the sample is non-random, France-centered, state-weighted, and thin in some sectors; all measures are self-reports; the comparison with scientists is descriptive rather than confirmatory; and respondents who are mandated to counter disinformation may overstate its prevalence or severity. Yet because these practitioners select detection instruments, draft regulations, and decide when to intervene, their perceptions are consequential. The study concludes that the people who shape and carry out democracies’ responses to disinformation are not scientists, they are civil servants, military officers, and regulators, and their beliefs, constraints, and reading habits are a crucial and understudied part of the problem. Understanding them, the authors argue, is a prerequisite for any effort to improve the response, and the current response, by their own account, is not enough.