According to Pew data based on a March 2025 survey, Fox News reaches 38% regular usage among American adults; 36% is not a general ceiling. With the generalization of digital access, recommendation algorithms play an increasingly important role in information exposure, alongside editorial filtering by newsrooms. The challenge is to understand what this fragmentation costs democratic deliberation, and what democracies that have preserved public anchoring do differently.
The Essential Points
- The common factual reference point that grounds democratic deliberation has vanished in the United States: no single source now commands more than a third of the audience.
- 86% of American adults get news online at least sometimes; 82% of journalists use at least one AI tool in their work (Pew Research Center, July 2026).
- Concern about uncontrolled AI in media rose from 18% to 26% in one year, signaling mistrust rising faster than adoption rates.
- Democracies with strong public broadcasting—BBC, ARD, France Télévisions, CBC—display informationally less fragmented spaces, according to NPR analysis from January 2026.
- Between 2028 and 2032, democracies will face a choice between two trajectories: strengthening common informational institutions or consolidating the archipelago of bubbles.
The Era of 36%: When No Single Voice Unifies
In March 2026, Pew reported that 36% of American adults typically turn first to their preferred news organization when breaking news occurs, according to a December 2025 survey; this figure does not measure an audience ceiling. To understand what this figure means, consider a baseline: during the 1960s-1970s, ABC, CBS, and NBC accounted for approximately 56% of households watching television on a typical evening; a figure of 75% should be understood as a share of televisions in use, depending on measurement context. The figure of 90% refers to the share of viewers present at 6:30 p.m., not all households. The audience was far from uniformly well-served by this system. But part of the audience shared a common foundation of facts.
That factual foundation can be less widely shared. The migration of information to digital platforms has unfolded over several decades and brought with it a multiplication of sources and intermediaries. But its effects accumulate. With the generalization of digital access, recommendation algorithms play an increasingly important role in information exposure, alongside editorial filtering by newsrooms. Recommendation systems on major platforms are principally designed to personalize content and satisfy users according to objectives specific to the platforms themselves; they are not, generally speaking, designed as institutions responsible for organizing a democratic public sphere.
Many recommendation algorithms use engagement and individual behavior signals, often combined with objectives of user satisfaction, retention, safety, and integrity.
Information practices and factual representations can vary according to sources and platforms used. The American information space is highly diversified, and source usage differs markedly across political groups.
AI in Newsrooms: Tool or Accelerant of Fragmentation
According to Muck Rack, 82% of journalists surveyed in its 2026 study use at least one AI tool in their work; this is neither a Pew statistic nor strictly a daily-use measurement for the United States alone. This figure deserves careful reading: it describes a transformation of journalistic production at a speed without parallel in the recent history of media.
AI in newsrooms can serve several functions. It automates low-value-added tasks—transcription, summarization, verification of structured data—and theoretically frees up time for reporting and investigation. Several major press groups, including The Associated Press and Reuters, have integrated these tools for coverage of financial markets and sports results for several years, with documented results in terms of production volume. The question of job stratification that results deserves separate attention, but the issue here is different.
AI poses a specific problem for the informational space when used to produce content at scale without rigorous editorial anchoring. NewsGuard has documented a rise in unreliable news sites that are predominantly or entirely AI-generated; the Columbia Journalism Review has covered certain cases and related work. Some AI-generated news sites publish at scale with minimal human editorial control and can disseminate low-quality or inaccurate content.
The signal of mistrust is eloquent. According to Muck Rack, journalist concern about uncontrolled AI use rose from 18% in 2025 to 26% in 2026. This movement generally precedes, in media history, a hardening of public expectations and then demands for regulation or certification. The acceleration of mistrust is a signal that institutional actors—regulators, press associations, platforms—would be wise not to ignore.
Public Broadcasting Stabilizes Where Market Alone Falls Short
International comparison is instructive, provided it is not made into an ideological argument. Public media can constitute widely-used sources of relative trustworthiness, without this sufficing to demonstrate a structurally less fragmented informational space. Comparison among these public media does not allow establishing that they maintain an audience of cross-reference within the population.
This audience does not mean uniformity. It means common anchoring. When a significant majority of citizens, even politically divided, have been exposed to the same facts verified by the same institution, the disagreement that follows is disagreement over interpretation and values, not over basic reality. This is precisely the kind of disagreement democratic deliberation is equipped to handle.
The American model relies on different architecture. NPR and PBS do not play the same institutional role as the BBC in the United Kingdom or ARD in Germany. This is a difference in design, not merely in funding.
This point merits nuance. Public broadcasting is not a panacea. The BBC is traversing a crisis of funding and trust, with progressive erosion of its audience among those under 35. France Télévisions struggles to find its place in digital usage. The public institution alone is insufficient: it must transform to remain relevant in a digital environment.
But its presence, imperfect, contested, in mutation, maintains an institutional weight that total absence of anchoring cannot simulate.
Media Ownership as Structural Variable
The Columbia Journalism Review has documented growing concentration of local media ownership in American hands of investment funds with short time horizons. Cost reductions and digital transformation have contributed to the decline of many local newsrooms and closures, but mechanisms and results differ across markets. In 2024, Medill documented 208 American counties without an established local news source and 1,563 with only one local source; the doubling between 2018 and 2024 is not established by this report.
The 2026 EY article addresses general media and entertainment trends; it does not document, in the verified text, this specific proposition about Meta, Google, and TikTok. The economic model of journalism financed by local advertising collapsed with the migration of classified ads to digital platforms. Many local media combine subscriptions or memberships, advertising, marketing services, philanthropy, and, for a significant portion of new digital entrants, nonprofit models.
This evolution is not irreversible. Documented experiments show alternative paths. The development of press cooperatives, a model studied by the Nieman Foundation at Harvard, produces stable local media in several mid-sized American cities. Nonprofit journalism foundations, modeled on ProPublica or the Texas Tribune, have demonstrated capacity to produce quality investigative journalism at a fraction of the cost of a commercial newsroom. These models do not replace mass media, but they constitute documented points of resistance against desertification.
Questions of public funding for local journalism are also posed differently depending on the country. Canada, for example, implemented tax credits in 2019 for subscriptions to local news media, a mechanism that stabilized dozens of regional outlets. This type of targeted intervention, neutral on content and focused on economic viability, is discussed in the United States without having found sufficient political majority.
Democratic Losses Tied to the Disappearance of a Common Factual Foundation
Democratic deliberation rests on a fundamental assumption: citizens debate on the basis of a sufficiently shared representation of reality. Disagreements concern values, priorities, and trade-offs. Not basic facts.
When representations of reality diverge, democratic deliberation can become more difficult. Elections can still function mechanically. But deliberation—discussion of collective choices, formation of informed preferences, the possibility of being convinced by argument—becomes difficult to make work. This tension between the epistemic conditions of democracy and the economic dynamics of progress runs through several contemporary debates.
The 2026 Edelman report notably highlights trust inequalities by income and increased insularity; it does not provide the comparison attributed to usual information sources. Trust levels can vary according to information practices. This divergence of trust can be associated with political polarization.
Daron Acemoglu, in his work on institutions and technology, poses the question of who captures the gains from technological progress and under what conditions those gains are shared. Applied to information, this question designates the beneficiaries of algorithmic disintermediation: platforms, whose market valuation depends directly on engagement volume, rather than citizens, whose capacity to deliberate depends on the quality of information they access.
Between 2028 and 2032, the Trajectories Taking Shape Now
The long-term evolution of the American information space, like that of many other democracies, remains uncertain. Two scenarios are taking shape, without anything making them inevitable.
The first extends current dynamics: consolidation of platforms, continuation of local media collapse, generalization of AI-generated content without editorial anchoring, acceleration of fragmentation. The rise in journalist concern about uncontrolled AI use can fuel worries about information trust, but it does not demonstrate widespread public disengagement from verified information. Democratic deliberation can become more difficult when communities rely on divergent information.
The second scenario starts from experiments already underway. Generalization of nonprofit models in local journalism, emergence of transparency norms on AI use in newsrooms, development of verification tools accessible to the public, and political decision to fund or protect common informational institutions constitute paths of action. This scenario does not restore the golden age of mass television, which was not without flaws, but it reconstitutes sufficient institutional weight for public debate to remain anchored in a shared factual reference point.
The signals that will make it possible to distinguish these trajectories are already readable. The speed at which American regulators—FCC, FTC, Congress—integrate informational stakes into their agenda is one. The choices major philanthropic foundations—Knight Foundation, MacArthur Foundation—make regarding their orientation toward structural support for local journalism is another. The capacity of platforms themselves to develop curation models that value editorial quality over engagement alone, under regulatory pressure or through anticipation, is a third.
Common informational institutions can help structure the informational space during the digital transition. A reformed BBC is better than no BBC; a strengthened NPR is better than an informational desert. The issue is institutional: certain collective architectures can contribute to democratic debate grounded in shared factual bases.
Answers to this question are being constructed now, before fragmentation becomes irreversible.
Sources
- Pew Research Center – Digital news consumption and AI in journalism, July 2026
- Columbia Journalism Review – analysis of media ownership and news deserts (no link: search “CJR news deserts 2024”)
- EY Media & Entertainment Trends 2026 (no link: report available on ey.com)
- NPR media analysis – international comparison of informational spaces, January 2026 (no link)
- Edelman Trust Barometer 2026 (no link: report available on edelman.com)
- Northwestern Medill Local News Initiative – data on American news deserts (no link: search “Medill Local News Initiative”)