A Funky Citizens analysis of 571.565 posts from the period 8-22 June 2026: domestic politics has generated almost three times more engagement than any other topic included in the analysis. It also was the target of the most extensive coordinated network included in the corpus.

Access the full report - HERE.

Key figures

• 12.4% of posts account for 27.8% of engagement. Domestic politics produced approximately 29.6 million of the ~106.7 million interactions in the corpus. Nearly three times more than the next highest-engagement topic.
• 612 average interactions per political post, compared to 126 for family and relationships content and just 72 for posts about tragedies and accidents.
• 402 accounts, 73,987 coordination events in a single active network, during the period June 10–22, 2026, explicitly targeting three names: Nicușor Dan, Adrian Veștea, and Ilie Bolojan.
• A single clickbait text about the "dismissal" of Nicușor Dan was distributed verbatim by 118 pages with a cumulative 91.6 million followers.
• 16 pages with religious and patriotic names ("Credință și Lumină," "România Populară," "Doar în România") distributed the same clickbait text about Ilie Bolojan identically, seconds apart from one another.

Why does this matter?

The period analysed caught Romania without a fully empowered government, nearly two months after the fall of the Bolojan Government and amid a succession of failed prime ministerial nominations. The country was operating under a real risk of snap elections, with approximately 10 billion euros from the PNRR (National Recovery and Resilience Plan) suspended until August 31, a budget deficit among the highest in the EU, and the leu at a historic low.

This is the second major governance crisis in less than two years, coming fewer than 18 months after the annulment of the 2024 presidential elections, officially justified by a coordinated amplification campaign on TikTok in favour of Călin Georgescu. This time, what can be documented is the behaviour: synchronisation and verbatim distribution.

What do the data show?

While the public was reacting disproportionately to political content, that very subject was the target of the largest coordinated network in the corpus. 402 Facebook accounts published, verbatim and at intervals of a few seconds, the same clickbait texts about negotiations for government formation. The pages distributing this content do not present themselves as political pages. They use names from the religious, patriotic, or entertainment sphere, with audiences built over time on innocuous content, then redirected toward narratives of delegitimisation, instability, and calls for investigations or dismissals.

The methodology, briefly

A corpus of 571,565 public Facebook posts, collected during the period June 8–22, 2026. 393,268 were thematically analysed after data cleaning. Of the 571,565 posts, 278,531 contain unique texts. Coordination was identified through a "sliding window" analysis: verbatim text distributed by different accounts within a maximum interval of 30 seconds. Full methodology details, interactive graphs, and all documented cases are available in the full report.

What circulates and how does it spread: an analysis of topics and coordination on Facebook during Romania's political crisis

Context: the second major crisis of trust in less than two years

This report analyses content on Romania's most popular social network — Facebook — during the period June 8–22, 2026. To understand why this report matters, however, one must go further back in time.

On November 24, 2024, Călin Georgescu, an independent candidate with pro-Russian sympathies and virtually unknown to the wider public just a few weeks earlier, surprisingly won the first round of the presidential elections with 23% of the vote, against a backdrop of strongly divided voting among the top 4–5 candidates.

In the days that followed, Romanian intelligence services declassified documents supporting the existence of a coordinated amplification campaign on TikTok and Telegram in his favour, involving accounts and influencers paid without declaring these sponsorships (approximately 80 dollars per 20,000 followers) — a pattern that the CSAT (Supreme Council of National Defence) described as "identical" to the campaigns used by Russia ahead of the invasion of Ukraine.

Pe 6 decembrie 2024, cu 48 de ore înainte de turul al doilea, Curtea On December 6, 2024, 48 hours before the second round, the Constitutional Court annulled the elections in their entirety: the first presidential ballot annulled in the history of the European Union for this type of interference. Months of protests, an institutional void, and a rewritten electoral calendar followed. The re-run elections of May 2025 brought Nicușor Dan to the presidency, with a narrow margin (53.6% versus 46.4%) over George Simion, the leader of AUR.

President Nicușor Dan promised a public report on the annulment of the elections at the time. That report does not exist yet.

What does exist, however, is a certainty — Romania emerged from that episode with a technically functioning democracy, but with institutional trust torn apart: part of the electorate accepted the annulment as a necessary measure in defence of sovereignty, while another part saw it as a blow struck by the very rule of law it claimed to defend. This fracture — not Russia or TikTok — is the real legacy of 2024: a society in which a significant portion of the population is prepared to believe that any unfavourable political outcome is, in fact, manipulated.

It is precisely that fracture that was exploited during the period we are analysing. On May 5, 2026, a motion of no confidence supported, unprecedentedly, by both PSD and AUR brought down the Bolojan Government. Nicușor Dan's first prime ministerial nomination, Eugen Tomac, failed (through the withdrawal of his mandate) in just ten days. The period June 8–22, 2026 — the subject of this report — caught Romania right in the middle of the next crisis: the second nomination, Adrian Veștea, was rejected by Parliament on June 22, with 189 votes out of the 233 required, AUR having left the chamber before the vote, and the candidate's own party, PNL, voting against him.

The country was operating without a fully empowered government, under a real risk of snap elections, with approximately 10 billion euros from the PNRR suspended (deadline: August 31), a budget deficit among the highest in the EU, and the leu at a historic low.

Why it matters: the publicwas reacting massively, and manipulation followed exactly that wave

Of the more than 571,565 Facebook posts collected during the analysed period, 393,268 were thematically analysed after the data cleaning process.

The most important finding does not concern the volume of posts, but the difference between volume and impact. The "Domestic Politics" theme accounted for only 12.4% of posts, yet generated 27.8% of total engagement — approximately 29.6 million interactions out of a total of around 106.7 million. This is nearly three times more than the next category by engagement, "Religion and Spirituality" (~11 million). A domestic politics post generated an average of 612 interactions, compared to approximately 126 for "Family and Relationships" and just 72 for "Tragedies and Accidents."

In other words, amid a political and institutional crisis, the public was reacting far more intensely to political content than to any other topic. This is natural — but it is also exactly the kind of fertile ground that coordinated networks seek out. The coordinated activity identified in the corpus was also concentrated on this very subject, with narratives constructed around the three central figures of the crisis: Nicușor Dan, Adrian Veștea, and Ilie Bolojan. The messages pursued recurring themes of delegitimisation, instability, and calls for dismissals or judicial investigations.

This is the central thesis of the report: the topic that attracted the most public attention during a political crisis was also the most intensely amplified through coordinated behaviours. This is happening fewer than two years after the first such episode, in a society whose trust in the democratic process is already at rock bottom. Instead of the lesson learned being one of greater transparency regarding these mechanisms, it appears that some actors have taken from the 2024 elections only the lesson of adapting their strategies and using the same type of mechanisms themselves.

Topic analysis

In order to analyse the content, posts were categorised into 29 topics: Family and Relationships, Tragedies and Accidents, Domestic Politics, Social and Community, Entertainment and Celebrities, Motivation and Personal Development, Religion and Spirituality, Sport, Crime and Violence, Health, Culture and Traditions, Economy, War and Geopolitics, Commercial Promotion, Justice and Corruption, Education, Environment, Governance, History and Identity, Moldova and Romania, Other, Security and Defence, Technology, EU and NATO, Elections and Parties, Censorship and Freedom of Expression, Russia and Ukraine, Geoeconomics and Trade, and Institutional Communication.

The full graph can be accessed HERE. 

What you will see: the percentage distribution of the 29 themes identified in the corpus, ordered in descending order by volume. The graph allows comparison of the weight of Domestic Politics with themes such as Religion and Spirituality, Social and Community, or Tragedies and Accidents. The graph is static and can be enlarged to read smaller labels.

The distribution of themes shows a corpus dominated by social, emotional, and popular-interest content. The largest categories are Family and Relationships, Tragedies and Accidents, and Domestic Politics, which together account for nearly 40% of all posts. By contrast, themes related to geopolitics, security, EU and NATO, or Russia and Ukraine have much smaller shares. This suggests that the dominant agenda of the corpus is oriented primarily toward domestic subjects, with a strong affective dimension.

In order to analyse how the public reacted to these themes, we used the "Total Engagement" indicator. This represents the sum of the principal forms of interaction with a post: reactions, comments, and shares.

The full graph can be accessed HERE.

What you will see: the same set of themes, this time ordered by total engagement, calculated as the sum of reactions, comments, and shares. Comparing this graph with the previous one, the differences become evident. Domestic Politics rises well above its share in volume, while Family and Relationships and Tragedies and Accidents fall significantly. The graph is static.

The most important observation is that Domestic Politics is a special case. Although it accounts for only 12.4% of the corpus, it generates 27.8% of total engagement. In other words, this topic produces far more interactions than its mere share of post numbers would suggest.

The situation is different for Family and Relationships and Tragedies and Accidents. Although they are the most widespread categories in the corpus, they generate proportionally less engagement. Family and Relationships accounts for 13.5% of the corpus but only 6.2% of total engagement, while Tragedies and Accidents accounts for 13.3% of the corpus and only 3.5% of engagement. On average, a post in these categories generates 125.7 and 71.9 interactions respectively.

There are also topics that, despite low volume, attract disproportionate interest. Religion and Spirituality accounts for 5.0% of the corpus but generates 10.3% of engagement. Motivation and Personal Development has a share of 5.8% in the corpus and 7.3% of engagement, while War and Geopolitics accounts for 2.6% of the corpus and 3.6% of engagement.

The Social and Community category is relatively balanced, with 9.0% of the corpus and 9.9% of engagement. By contrast, Entertainment and Celebrities has a share of 7.7% in the corpus but generates only 5.6% of engagement.

Even among very small themes, categories appear that can generate high levels of interest when posted. Elections and Parties accounts for only 0.2% of the corpus, while Russia and Ukraine accounts for only 0.1%. Although their contribution to total volume is small, these subjects can produce high levels of engagement.

The conclusion is that post volume and capacity to generate engagement do not overlap. Some themes are very present in the corpus but produce relatively few interactions. Others — in particular Domestic Politics, Religion and Spirituality, and Motivation and Personal Development — mobilise the public to a far greater extent than their frequency would suggest.

Possible explanations: what the data suggests and what remains hypothetical

The data do not permit the direct attribution of intent or a "sponsor," but they do support the formulation of hypotheses that can be tested.

The amplification of negative messages about Nicușor Dan, Adrian Veștea, and Ilie Bolojan is concentrated precisely during the period of negotiations for government formation. This pattern is compatible with an attempt to consolidate the perception that the governance process is unstable and lacking in legitimacy. This narrative, regardless of source or of attacks between the political actors involved, also serves those parties interested in making the current institutional and political framework appear beyond saving.

At the same time, the repeated calls for judicial investigations and dismissals, distributed verbatim by dozens of accounts within intervals of just a few seconds, indicate coordinated behaviour that exceeds the patterns of ordinary political debate. Viewed together with the temporal synchronisation, the repetition of the same narratives, and the systematic disguising of pages carrying political messages under religious, patriotic, or entertainment identities, this behaviour is compatible with an objective of amplifying the perception of political instability.

A second hypothesis, equally plausible, concerns normalisation: after 2024, the use of coordinated networks to influence public perception of a political process is no longer an exotic instrument reserved for foreign intelligence services. The model has been documented publicly, in detail, in the press and in official reports from various states interested in this phenomenon (though not in Romanian ones). It is possible that political actors, campaign consultants, or simply domestic opportunists have absorbed the lesson — not the specific content, but the fact that the mechanism works, that it is cheap to replicate, and that, for the time being, it carries no visible consequences.

We cannot establish with certainty which of these explanations is correct, or whether both coexist. What we can establish is that the artificial amplification infrastructure identified in this corpus did not appear from nowhere, and the fact that it targets with precision the actors of a governance crisis, in the tone and vocabulary typical of the electoral disinformation of 2024, does not appear coincidental.

From TikTok to Facebook: same infrastructure, different players

The comparison with 2024 is instructive not because the two episodes are identical, but precisely because they are not — and the differences show how far the mechanism has become commonplace. In 2024, the official investigation documented direct payments to influencers, coordination on Telegram among those organising the campaign, unified hashtags, and a single platform — TikTok — chosen precisely for its influence over a young audience and for the opacity of its recommendation algorithm.

In 2026, the mechanism is less technically sophisticated, based on verbatim text distributed by Facebook accounts, without evidence of payment or coordination on an external channel, but equally effective: pages with religious, patriotic, or entertainment names, with audiences of tens or hundreds of thousands of followers, built over time on innocuous content, redirected within seconds toward political narratives.

What remains identical is the basic idea: the public does not see a propaganda network, but a page it already trusts: "Amintiri frumoase" (Beautiful memories), "PtSuflet" (For the soul), "Credință și Lumină" (Faith and light.) The difference is that in 2024 the presumed authors were external, and the institutional response was commensurate: the annulment of a presidential election. In 2026, if the authors are domestic — as the absence of any demonstrable link to a foreign actor suggests — the institutional response so far has been equally clear: zero.

How it works: the mechanics of coordination

The basic infrastructure

Of the 571,565 posts analysed, 278,531 contain unique texts.

To identify coordinated behaviours, we used a "sliding window" analysis. We tracked posts with identical text distributed by at least 50 different accounts, within a maximum interval of 30 seconds, across the entire period June 8–22, 2026. Each time two accounts published the same text within this time window, a coordination event was recorded. All these events were then aggregated into a network graph, in which the links between accounts are weighted according to the frequency of coordination.

What you will see: a table of the most frequently identically distributed texts by multiple accounts and the number of accounts that published each text.

The analysis identified numerous instances in which different accounts — including accounts with the same name or very similar names but different platform identifiers — repeatedly published the same verbatim texts simultaneously or within an interval of less than 30 seconds. This pattern constitutes an indicator of coordinated behaviour.

Domestic politics, in a coordinated rhythm

An interesting case is that of the sensationalist text "Ludovic Orban announces THE END of Nicușor Dan! THE DISMISSAL of the president will… See more," which was disseminated on June 22, 2026 in several short coordination waves, in which different pages published the same message at intervals of a few seconds. The first episode appears in the morning, between 12:02:52 and 12:03:13, through the pages Case de Vis, femina.ro, and ROL.ro, quickly generating 208 likes, and distribution then intensifies around 14:40–14:43, when the text is successively picked up by networks of pages with a generalist, emotional, entertainment, or community profile, such as Observatorul, România Noastră, DRAGOSTEA MEA, România ta, Amintiri frumoase, PtSuflet, and Fabricat în România.

The largest event in this sequence is that of 14:42:09–14:42:39, with 11 posts and 89 likes. A new wave appears at 15:26–15:27, with Facebook pages such as a1.ro, DOZA De RAS, DRAGOSTEA MEA, PtSuflet, and Terapie prin zâmbet.

The coordination events span a period of nine days, from June 22, 2026 at 12:02 to June 30, 2026 at 14:40, with 118 distinct pages having published the text verbatim. Chronologically, activity is not uniform but appears in concentrated waves: the first and most extensive wave takes place starting June 22, between 12:02 and 00:02 Bucharest time, when 77 posts are published; this is followed by a small peak of just 4 posts on June 23, around the afternoon, between 15:00 and 15:07, then 10 posts on June 24, between morning and evening.

Activity enters an upward trend again on June 25, with 37 posts between 16:40 and 21:22, drops to 20 posts on June 26 between 08:30 and 20:00, then rises again to 32 posts on June 27 between 07:19 and 18:30, and to the most intense second wave of 59 posts on June 28, distributed between 09:15 and 20:51. The last notable impulses appear on June 29, with 40 posts concentrated in a very short interval between 06:40 and 07:24 in the morning — indicating near-simultaneous publication from different accounts — and the event closes with a single isolated post on June 30 at 14:40.

The full graph can be accessed HERE.

From the perspective of engagement metrics, the total sum of likes recorded across the 280 posts is 1,804, shares 173, and comments 2,405, while the number of views (ViewsCount) reported directly at post level is 0 for all records. However, at the level of the publishing pages, audience data (ChannelEngagement) reveal a network with far greater exposure potential: the cumulative number of followers of the 118 pages involved exceeds 91.6 million, and the total number of views reported at channel level reaches approximately 233.7 million, with ROL.ro being the largest, with 2,146,908 followers.

This discrepancy between the low engagement actually generated by each post (likes and comments numbering in single or double digits) and the very large collective audience of the pages suggests a typical coordinated amplification pattern, in which an identical message is rapidly distributed through an extensive network of high-reach accounts, rather than through organic virality based on public interaction.

The majority of verbatim-distributed texts in the "Domestic Politics" category are clickbait posts constructed around political actors such as Nicușor Dan, Ilie Bolojan, Adrian Veștea, Eugen Tomac, Sorin Grindeanu, Victor Ponta, Călin Georgescu, and the main parliamentary parties: PSD, PNL, USR, and AUR.

Recurring subjects include negotiations for government formation, prime ministerial nominations, resignations, suspensions, inter-party conflicts, and alleged "revelations" about political leaders. The language is consistently sensationalist and uses expressions such as "shock," "alert," "total detonation," "reversal of the situation," "the end," or "fatal blow."

The texts offer only a fragment of information and direct the reader to "the first comment," where the link is placed. Formulations such as "major political scandal," "the political scene is shaking," or "crushing announcement" are used to create the impression of a spectacular development and to encourage clicking the link.

The majority of narratives centre on Ilie Bolojan. The texts promote successive and often contradictory scenarios: that he has yielded in negotiations, that he is set to remain in office, that he could be nominated again after Adrian Veștea's withdrawal, that he is under attack from Alina Gorghiu, contested by PSD, or targeted by alleged accusations or "recordings" that Nicușor Dan would make public.

Eugen Tomac also appears frequently in these messages. The texts alternately claim that he has been proposed or nominated as prime minister, explain why he would have accepted the position, suggest that USR and Dominic Fritz are deciding the fate of a supposed "Tomac Government," or present him as attributing to Ilie Bolojan the responsibility for the political deadlock.

In the case of Adrian Veștea, the posts present him as the prime ministerial nominee of Nicușor Dan and speculate about a possible "Veștea Government." At the same time, they promote attacks on him and information about the prospective cabinet, his assets, family, or a possible withdrawal from the race for the prime ministerial position.

The interactive CIB network graph can be accessed HERE.

What you will see: an interactive graph in which each node represents a Facebook account, and the links between nodes represent texts distributed verbatim within an interval of less than 30 seconds. The thicker a link, the more identical texts the two accounts have shared. You can click on any node to see the account name and number of coordinations, move nodes to visually reorganise the network, and use the zoom function to explore the densest areas. The five nodes described below are visibly larger than the rest. 

The political coordination network identified comprises 402 nodes and 73,987 coordination events during the period June 10–22, 2026. The most active pages are "Fericire Si Speranță Ingerii Inimii Tale" with 456 coordinations, "Pagina iubitorilor de Dumnezeu-Tu, il iubesti?" with 441, "DOZA De RAS" with 427, "Amintiri frumoase" with 415, and "PtSuflet" with 411.

All five pages display the same pattern of activity. Each publishes exactly 17 posts, using 15 text variants, with nearly identical temporal synchronisation and the same main targets: Nicușor Dan, Adrian Veștea, and Ilie Bolojan. Although they represent only 2.9% of the total nodes in the network, the uniformity of their behaviour suggests the existence of a common coordination mechanism.

The amplified narratives follow three recurring directions: the delegitimisation of political actors through attacks on their wealth, competence, or character; the accentuation of the perception of instability through references to crises, tensions, resignations, and betrayals; and calls to action, such as demands for judicial investigations, dismissals, or suspensions.

The language is consistently sensationalist and uses formulas such as "shock," "alert," "total detonation," "reversal of the situation," or "fatal blow." The texts are incomplete and frequently end with formulas such as "continued in the first comment," designed to lead users to access the distributed link.

An important detail: the names of the pages do not reflect their content 

The pages that coordinately distribute political content do not present themselves as political pages. Their names are inspired by religious, patriotic, emotional, or entertainment themes, such as "Credință și Lumină," "România Creștină," "Amintiri frumoase," "Dumnezeu e bun," "PtSuflet," or "DOZA De RAS." Other pages use names suggesting news sources, such as "News Romania," "Stiri reale," or "Curiozități Ascunse."

The contrast between the displayed identity and the distributed content is significant. Political messages are published through pages that initially attract audiences interested in religion, nostalgia, entertainment, patriotism, or viral news — not political debate.

Overall, the network consists largely of pages with an apparently neutral or apolitical identity, used for the coordinated distribution of messages about the political crisis, the actors involved, and institutional tensions. This pattern suggests that audiences built around non-political themes can subsequently be repurposed for the amplification of political content.

Motivation and personal development: same technique, different content

The "Motivation and Personal Development" category is dominated by clickbait content: jokes, marital humour, horoscopes, and anecdotes whose continuation is moved to "the first comment." Coordinated behaviour also appears in this category, though in a less complex form than in the case of political content.

The network comprises 138 nodes and 5,474 coordinations during the period June 9–20, 2026. The most active pages are "Să râdem împreună" with 61 coordinations, "Tezaur de Curiozități" with 60, "Sufletel" and "Sunt frumoasă așa cum sunt" with 57 each, and "Trandafirul albastru" with 55.

All these pages display a similar pattern of activity: seven posts and seven text variants distributed in a coordinated manner. Compared to the political network, the structure is less rigid and does not target specific political figures or persons.

The content is heterogeneous and includes horoscopes, jokes, and motivational messages. This suggests that the main objective is attracting and retaining public attention, rather than promoting a political narrative. However, the distribution mechanism remains the same: apparently innocuous pages accumulate visibility and audience by publishing nearly identical texts in a coordinated manner.

From this perspective, the network is relevant not through the topic addressed, but through the infrastructure it builds. Audiences created around apparently neutral subjects can subsequently be repurposed for distributing other types of content, including political content*.

The interactive graph can be accessed HERE.

What you will see: an interactive graph constructed using the same methodology as for Domestic Politics, but with visibly lower density. The network is more dispersed and does not feature dominant nodes comparable to those identified in the political network. How to interact: click on nodes to see page names; use the zoom function to explore the links between them.

At first glance, this theme may seem less important than the one related to domestic politics. However, the analysis shows that coordination mechanisms are not used exclusively for political content. Some networks appear to gradually build audience and visibility through apparently innocuous content, subsequently using the same infrastructure for the coordinated distribution of other types of messages.

From this perspective, the network's relevance lies not in the topic addressed, but in the coordinated behaviour it highlights and in the fact that the same infrastructure can be repurposed in different contexts.

Other networks, by area of interest

On the topic of Health, 84 accounts coordinately distribute content combining alarmist medical themes (stroke symptoms, colon cancer) with wellness, horoscopes, and curiosities. The names of the accounts with the most matches include "Familia Becali," "George Becali," "Fericire Si Speranță Ingerii Inimii Tale," "Cinematografia Veche," "Observatorul," "DRAGOSTEA MEA," "Pentru Tine," "DOZA De RAS," and "Românii din Străinătate."

A concrete case is the page "Unofficial: Andra Măruță," which has over 400,000 followers, created in March 2014 as "Johny Bravo Romania," renamed in 2017 to "Iubim Crăciunul," and since May 2017 using the celebrity's image to disseminate disinformation at fixed hours.

The interactive graph can be accessed HERE.

What you will see: the same type of interactive graph as for domestic politics and health, but with visibly lower density. The network is more dispersed, without clear dominant nodes. How to interact: click on nodes to see the page name; use zoom to view the links.

The "Justice and Corruption" category in the corpus is dominated by judicial clickbait and pseudo-breaking news: references to the DNA (National Anticorruption Directorate), persons "caught," investigations, celebrities, or politicians invoked in incomplete headlines, with the ending pushed toward "see more" or toward comments. The network comprises 104 nodes with 8,542 coordinations.

The five main nodes are: 

• „Natura cu frumusețile ei” (Nature and its beauties) with 231 coordinations;
• „Fericire Si Speranță Ingerii Inimii Tale” (Happiness and hope, the angels of your heart) with 228 coordinations; 
• „Pagina iubitorilor de Dumnezeu-Tu, il iubesti?” (The page of God lovers. Do you love Him?) with 228 coordinations; 
• „Ingerii Inimii TALE” (The angels of YOUR heart) with 217 coordinations;
• „PtSuflet” (For the soul) with 216 coordinations.

The structure shows that the justice theme is used more as a pretext for capturing attention than as a domain of genuine civic information. The coordinated texts include formulas such as "Ciprian Ciucu, the Mayor of the Capital, urgently summoned to the DNA," "Denise Rifai has arrived at the DNA," "HE HAS BEEN CAUGHT! The most wanted man in Romania," or "The theatrical coup by Ilie Bolojan," mixed with tabloid and emotional content. The network's relevance lies in the fact that apparently innocuous pages — religious, motivational, entertainment, or community-oriented — synchronously distribute content with the appearance of judicial news, consolidating an amplification mechanism based on suspense, distrust, and clickbait, not on verifiable information.

The interactive graph can be accessed HERE.

What you will see: the same type of interactive graph depicting a network fragmented into several small clusters rather than a single central nucleus. Coordination appears across separate groups of generalist/emotional pages that redistribute the same few clickbait texts, suggesting a modular amplification infrastructure with parallel nuclei rather than a single, very dense network.

The unexpected link: Bolojan, faith and clickbait

16 Facebook pages distributed, seconds apart from one another, the same clickbait-formatted text: "BREAKING NEWS! Bolojan has… See more" — a type of headline that links to the first comment and leaves the rest to the imagination. The mechanism exploits a documented phenomenon: the mind automatically fills in missing information.

The most active nodes are pages with patriotic, religious, or generalist names: "Romania Populară," "Romania Iubeste," "Romania Fantastica," "Credință și Lumină," "Doar in Romania," "Femei Libere." The recipe is simple: populate accounts tailored to the preferences of multiple audiences, keep them active with periodic sensationalist posts, and use them when needed for political messages.

The interactive graph can be accessed HERE.

What you will see: a graph showing the 16 pages that distributed the same clickbait text about Bolojan seconds apart. How to interact: click on nodes to see the page name; note that all nodes are connected through a single text.

Overall, this phenomenon confirms the observations made thus far: apparently apolitical or general-interest pages are used for the repeated amplification of messages that bombard and intoxicate us in our moments of greatest societal vulnerability.

Methodology

Corpus analysed: 571,565 Facebook posts, period June 8–22, 2026. After data cleaning, 393,268 posts were thematically analysed and categorised into 29 thematic clusters.

Thematic analysis used a Machine Learning (ML) model for grouping similar texts, interpreted through a hybrid Human-In-The-Loop system based on NLP (Natural Language Processing) and RAG (Retrieval-Augmented Generation).

Coordination detection uses a rolling event-based sliding window. Posts were grouped by normalised verbatim text and ordered chronologically. Each post was compared with previous posts containing the same text, within a retrospective window of maximum 30 seconds. If the previous post came from a different account, a coordination pair event was recorded. Events were aggregated into weighted edges, and the edge weight represents the total number of co-occurrences; unique texts represent the number of distinct verbatim texts involved.

In lieu of a conclusion

Networks of coordinated accounts can distort the perception of the real level of support enjoyed by political actors and can make it more difficult to differentiate between authentic participation and coordinated activity. In some cases, they use without consent the photographs of real individuals to construct false identities and increase the apparent credibility of accounts.

The cases documented in this report do not appear to be isolated incidents — and, equally importantly, they do not appear to be the first episode of this kind in Romania's recent history. Fewer than two years after a presidential election annulled for coordinated amplification in favour of a candidate, the country is confronted with a new episode of coordinated manipulation of public discourse, this time without an identifiable external source and without any visible institutional consequence. Regardless of whether the purpose is attracting traffic through clickbait, amplifying political narratives, or influencing public conversation during periods of crisis, the observed mechanism remains the same: networks of accounts that coordinately distribute the same content to create the impression of greater support or interest than results from authentic user activity.

From this perspective, the principal risk is represented not only by the individual content of the posts, but by the infrastructure that makes their coordinated amplification possible — an infrastructure that, as this report shows, survives changes in political context, platform, and possibly even in the party commissioning it.

The fact that some of the networks identified have remained active for months without visible intervention from the platform raises legitimate questions regarding compliance with the obligations set out in Article 35(1) of the EU Digital Services Act (DSA), which requires very large online platforms to implement effective measures for limiting coordinated inauthentic behaviour.

If the annulment of a presidential election was not sufficient to prompt a systemic response from platforms and regulatory authorities, it remains to be seen what threshold would be necessary for this type of behaviour to be treated as a structural problem rather than a series of separate incidents.

This material is funded through the Tech Accountability Grants 2026 Programme of the European Fact-Checking Standards Network (EFCSN), under the CERV grant (no. 101236606) awarded by EACEA.

The views and opinions expressed in this material belong exclusively to the authors and do not necessarily reflect the position of the European Union or the European Education and Culture Executive Agency (EACEA). Neither the European Union nor the funding authority can be held responsible for them.