In December 2024, Romania became the first EU country to annul a presidential election over a social-media disinformation campaign. Six months later, researchers asked Romanians how much they trust their news. The answer: 26% — lower than 44 of the 48 markets measured on Earth. This is the story of a trust recession — told entirely in published numbers.
Every year, the Reuters Institute at Oxford asks nearly 100,000 people in 48 markets the same blunt question: can you trust most news, most of the time? The global answer has been stuck at 40% for three years. The national answers are where the story lives.
At the top of the ledger sits Finland, at 67% — a society where two out of three people still extend their news the benefit of the doubt. The Nordic and Atlantic democracies cluster behind it. Then the line slides south and east, past a bruised Britain at 35%, down to the bottom of the continent: Hungary and Greece, at 22% — the lowest trust in the entire 48-market sample.
And just above them, at 26%: Romania. Rank 44 of 48, down one point in a single year. Roughly one Romanian in four believes the news. Three in four start from doubt.
One Romanian in four believes the news. Three in four start from doubt.
It would be comfortable to read this as cynicism — a national habit of suspicion. The harder reading is that the doubt is earned. The same report documents Romanian television stations charging presidential candidates up to €200,000 per broadcast for paid electoral programmes made by the stations' own journalists — legally, under Romania's electoral rules. When news can be bought, doubting it isn't a pathology. It's pattern recognition.
But a society cannot run on doubt alone. Trust doesn't vanish — it migrates. And in November 2024, Europe got to watch, in real time, exactly where Romanian trust had migrated to.
What follows is not commentary. It is a case file — the documented sequence of events that made Romania the world's reference case for what happens when an information ecosystem fails. Every entry below is sourced.
This is the part the international coverage kept missing. December was not a story about one candidate, one platform, or even one hostile state. It was a stress test of a national listening infrastructure — the layer of institutions, newsrooms and habits through which a society hears itself. The test found the infrastructure already hollow. The disinformation campaign didn't create the vacuum of trust. It moved into a vacancy.
Disinformation is opportunistic. It flows where resistance is lowest. Two independent European measurements — one of skills, one of societal resilience — show where the gates stood open.
The first is brutally simple. Eurostat measures the share of people aged 16–74 with at least basic digital skills — the ability to find, evaluate and manage information online. In 2023, the EU average was 56%. The Netherlands reached 83%. Romania recorded 28% — the lowest in the European Union.
Now look at the same gap by generation, because this is where the comfortable story collapses. The comfortable story says: the young are digital natives; time will fix this. The data says otherwise. Among Romanians aged 16–24 — the TikTok generation itself — only 47% reach basic digital competence, against an EU average of 70%. Growing up inside the feed is not the same as being able to read it.
Growing up inside the feed is not the same as being able to read it.
The second measurement is broader. Since 2017, the Open Society Institute – Sofia has published the Media Literacy Index — a composite of media freedom, education quality (weighted heavily toward PISA reading literacy), interpersonal trust and e-participation. It doesn't measure disinformation. It measures a society's predicted resilience to it.
Romania sits in the index's fourth cluster of five — the band of societies the model flags as structurally vulnerable: weaker media freedom, education deficits concentrated in reading literacy, and thin interpersonal trust. If you read Story 01 of this series, that last phrase should sound familiar. The narrow circle of trust we mapped in the values data is the same soil in which the December campaign grew. The two stories are one story.
Low trust doesn't mean people stop consuming information. It means they change how they consume it — and what they're willing to give in return. Four Romanian numbers, side by side, draw the new contract.
Source: Reuters Institute Digital News Report 2025 — Romania country page. All values verbatim.
Read together: a quarter of the country gets news from an entertainment algorithm, a quarter actively redistributes news through private channels, barely one in ten pays for any of it — and only one in four believes what arrives. Information still circulates at full volume. What's gone is the willingness to underwrite it — with money, or with belief.
The region-wide symptom is avoidance. Across all 48 markets, 40% of people now sometimes or actively avoid the news — and the European champions of looking away are all in Romania's neighbourhood: Bulgaria at 63%, Turkey and Croatia at 61%, Greece at 60%. The Media Literacy Index authors cross-checked their 2026 scores against this avoidance data and report a clear inverse pattern: the more vulnerable a society is to disinformation, the more its citizens turn away from news altogether. Doubt, taken daily, curdles into exhaustion.
And when people in low-trust environments do try to verify something, what do they reach for? Globally, the most common answer is still "a news source I trust" (38%), then official sources, then fact-checkers (25%). At the very bottom of the verification toolkit: AI chatbots, at 9%. Even in the wreckage of institutional trust, people still want a someone — not a feed, not a model — to vouch for reality.
People don't want a feed to vouch for reality. They want a someone.
Which brings us to the one number in this story that points upward. Between its 2023 and 2026 editions, the Media Literacy Index recorded something rare for this part of the map: Romania improved — from 32 to 37 points, climbing three places between the two editions, while several of its cluster neighbours stood still or fell. And inside the Reuters data, two donation-funded investigative newsrooms, Recorder and G4Media, appear among Romania's most-used online news brands — outlets whose entire business model is the voluntary underwriting of trust. Small numbers. Real direction.
Roads carry goods. Grids carry power. Trust carries meaning — and like any infrastructure, it is invisible until it fails, expensive to rebuild, and impossible to import.
The data in this story describes a country whose meaning-infrastructure failed a live stress test in December 2024 — and then did something unusual: it noticed. A court ruled. A measurement moved. Two newsrooms funded by their own readers stand among the country's most-used news brands. None of this guarantees anything. All of it is real.
This series exists because I believe the first repair is the cheapest one: listening carefully, in public, with sources shown. Not louder claims — better ledgers. Every number above is published, named, and linked below, so you can doubt me properly. That, in the end, is the whole proposal: doubt that has somewhere to go is called verification. Doubt with nowhere to go is called despair. The difference between them is infrastructure.
— END · STORY 02 —
This essay is a narrative reading of published research — not original statistical work. Unlike Story 01, which mixed verbatim figures with illustrative comparative positions, every numeric value charted or cited in this story is a published figure, reproduced verbatim from the source named beneath it. No values were estimated, interpolated or illustrated.
1. Trust, consumption and the Romanian case file (Charts 01, stat strip, timeline): Reuters Institute Digital News Report 2025 (University of Oxford; n≈100,000 across 48 markets, ~2,000 per market, fielded January–February 2025 by YouGov). Country trust values: Finland 67, Denmark 56, Norway 54, Portugal 54, Sweden 53, Ireland 51, Netherlands 50, Poland 47, Switzerland 46, Germany 45, Belgium 43, UK 35, Romania 26, Hungary 22, Greece 22; global average 40. Romania page authored by Raluca-Nicoleta Radu (University of Bucharest): trust 26% (rank 44/48, −1), pay for online news 12%, TikTok as news source 25%, sharing 24%, market figures (€1bn revenue, €778m advertising, €180m state funds, €150m party communication spending, €200,000 paid broadcasts, €1m influencer payments), and the election chronology. Verification behaviours (trusted source 38%, fact-checkers 25%, AI chatbots 9%) and news avoidance figures (global 40%; Bulgaria 63%, Turkey 61%, Croatia 61%, Greece 60%) from the DNR 2025 executive summary. Methodological caveat: DNR samples represent online populations and are fielded via online panels; in countries with lower internet penetration the sample skews more connected and educated than the general population. Romania's internet penetration: 89%.
2. The ~25,000 TikTok accounts figure originates in declassified Romanian intelligence (SRI) material released in December 2024 and is reported via the Foreign Policy Centre's "Networks of Influence" (2024) and international press. It is an intelligence estimate, not an audited count, and is labelled as such in the timeline.
3. Digital skills (Charts 02–03): Eurostat, dataset isoc_sk_dskl_i21, 2023 reference year (news release 15 Dec 2023): EU 56%, Netherlands 83%, Finland 82%, Denmark 70%, Poland 44%, Bulgaria 36%, Romania 28% (precisely 27.7%). Age splits from the European Commission's Digital Decade 2025 Country Report — Romania: 16–24: 47.19% vs EU 69.98%; 65–74: 6.17% vs EU 28.19%. Eurostat's 2025-reference-year release (April 2026) shows Romania at 31.8%, still 27th of 27 — noted in the chart source line. The 80% marker is the EU's official Digital Decade 2030 target.
4. Media Literacy Index (Chart 04): Open Society Institute – Sofia, MLI 2026 (policy brief, January 2026). The index is a predictor model of societal resilience (media freedom 40%, education 45% — chiefly PISA reading, interpersonal trust 10%, e-participation 5%), not a direct measurement of disinformation exposure; its authors quote Box's aphorism that "all models are wrong, but some are useful," and so do I. All charted scores verbatim from the official table: Denmark/Finland/Ireland/Netherlands 71, Norway 70, Estonia 69, Sweden 68, Germany 62, UK 62, France 55, Poland 54, Italy 51, Croatia 47, Hungary 43, Ukraine 38, Greece 37, Romania 37 (rank 31/41, cluster 4), Serbia 35, Bulgaria 32, Turkey 30. Romania's change vs the 2023 edition (score 32, rank 34) is taken from the report's own comparison table: +5 points, +3 positions; statements about neighbouring countries' movement refer to the same table.
5. Interpretive frame: The connective tissue — "trust migrates," "listening infrastructure," the link to Story 01's interpersonal-trust findings — is editorial interpretation and is written so that it cannot be mistaken for a survey finding. Where the chapter on December describes events, each timeline entry carries its own source tag.