Social media usage and Armenia’s 2026 election

On June 7, Armenians voted in a parliamentary election that returned Nikol Pashinyan’s Civil Contract to power with a little less than half of all votes. This election, however, was not only fought in the traditional sense; it was also fought in an online information environment shaped by disinformation, diasporic relations, and a history of censorship.

Seventy years of Soviet state media left Armenians in a difficult reality. In 2001, 96.8% of the population reported having little trust in mass media, indicating a deep-seated suspicion of official information sources. Although these numbers have shifted over the last 25 years, the willingness of the Armenian people to turn to alternative media has consistently reflected this distrust. By 2024, around two-thirds of Armenians utilised social media as a primary news outlet; however, this figure is somewhat misleading, as traditional media still dominates those above 45 and in rural populations. Social media has taken a unique position in Armenia, not by supplementing already well-established journalism, but by filling a vacuum.

It was in 2018 that social media first became a well-utilised tool in Armenian politics. Pashinyan’s strategy of bypassing captured broadcast media relied on multiple new media platforms, each with distinct functionalities: Facebook Live as a broadcast infrastructure, Telegram as a closed coordination tool, and livestreaming as a real-time accountability mechanism that made violence against protesters instantly costly and visible. With traditional media aligned with the ruling party, reporters had to follow Facebook groups and Telegram conversations to find out where protests would be held and what ideas would circulate; the distinction between online platforms and the press had effectively disappeared. Crucially, though, the underlying media structure that had produced this reality did not change; only the government had. Public distrust in the media continued, and Armenia’s information environment remained susceptible to manipulation.

That underlying distrust carried into the 2021 elections, where social and new media again proved a valuable political tool, although this time the strategies initially utilised to mobilise were instead deployed to polarise. The same platforms that had enabled horizontal civic communication in 2018 became vectors for blame, conspiracy, and grief, a shift shaped in large part by the trauma of the Second Nagorno-Karabakh War. Telegram underwent a key infrastructural shift in this regard: unlike Facebook, which is more easily monitorable, Telegram’s closed channel architecture made coordinated disinformation almost impossible to track in real time. Most political groups in Armenia effectively established a communication strategy built on emotional registers of betrayal, loss, and national humiliation; registers that would come to dominate short-form political content five years later.

By 2026, Armenia’s information environment remained shaped by these accumulated realities. Over the last 5 years, social media usage has only become more diverse, influenced by a culmination of factors such as age, gender, interests, and socioeconomic class. Social media usage had grown more diverse, influenced by age, gender, interests, and socioeconomic class, and a new set of tools had entered the picture. Short-form content became a major player in political communication, with TikTok, YouTube, and Instagram all serving as platforms for fast-paced content well suited to the dissemination of disinformation. An analysis by Respense examined around 57,561 media mentions across websites, Telegram, Facebook, TikTok, Instagram, X, and broadcast television; a reflection of Armenia’s pluralistic yet polarized information space. Social media was utilized by both major parties in the run-up to the election, creating a competing reality where outlets were instrumentalised to amplify mutual insults and divisive campaign rhetoric, offering little substantive analysis of policy platforms. Notably, around 17.5% of all TikTok videos analyzed were labeled high risk for disinformation.

The disinformation that circulated reflected strategies well-known and well-tested around the world. Themes centred on emotional pulls familiar from 2021: betrayal, loss, and national humiliation. These themes were built into fear-driven narratives designed to exploit public anxieties surrounding peace efforts and the existential threat of renewed conflict. The data reflects this starkly: the two primary opposition narratives were ‘the government betrayed Nagorno-Karabakh’, generating 1,230 videos and 15.9 million views, and ‘the 2026 elections will be rigged’, generating 594 videos and 10 million views. It should be noted that Pashinyan remains the political actor in Armenia with the largest single online following, though the opposition collectively pulled in approximately twice the amount of views. Pashinyan’s own narratives followed a similar emotional logic, engaging public anxieties surrounding peace efforts by warning that a war with Azerbaijan was imminent should the opposition win.

Alongside emotionally driven narratives came the use of AI-generated multimedia and deepfakes: fabricated clips designed to look like news broadcasts, showing falsified documents to discredit political candidates and front pages of well-known international newspapers impersonated to lend false credibility to fabricated stories. A further identifiable strategy was the use of foreign bots and influence networks, placing false information on foreign platforms and then legitimizing and spreading it through official and unofficial channels of regional actors. Many domestic online spaces also exhibited a pay-to-play dynamic, with political entities buying manipulated digital visibility and sponsoring pages to push targeted attacks on opponents. Social media was no longer simply a space for communication and mobilization; it was used simultaneously to mobilize, legitimize, disinform, and suppress

To understand the scale of the problem, it is worth mapping the specific clusters of disinformation that circulated during the campaign. The content was not random; it was calculated and drawn on deep wells of public anxiety that have been building in Armenia for years. The dominant cluster was security-based, with the most widely circulated narratives portraying Armenia on a path toward military confrontation, with the drawing of comparisons to Ukraine’s lived experience. These narratives were frequently delivered through fabricated news content. Hundreds of fake videos had been published by early May 2026 alone, including fabricated clips falsely claiming that NATO instructors were present in Armenia and that a military conflict with Russia would be provoked after the election. Security fears were especially potent given Armenia’s lived memory of violence, and disinformation consistently exploited this wound. The second major cluster concerned Nagorno-Karabakh itself, with disinformation narratives spreading falsities, especially among the Armenian diasporas in Russia, blaming the current government for the consequences derived from the conflict. An inherent attempt to frame the conflict as an emotional binary reality, removing the complex regional context that exists.

The third cluster focused on economic realities, with false narratives about the economy being prevalent throughout the campaigns. This included fabricated claims about what EU integration entails and what it would mean for Armenian households, jobs, and trade. These were reinforced by real economic pressures, as in late May 2026, Russia’s consumer protection agency temporarily suspended imports and added restrictions on Armenian flower exports to Russia; a move perfectly timed to coincide with the electoral campaign, recalling the economic pressure Russia had applied to Moldova and Georgia when those countries pursued European integration.

A fourth cluster operated on cultural and identity lines, linking geopolitical messaging in order to reach audiences otherwise indifferent to foreign policy debates. This included fabricated claims that EU integration carried a mandatory condition to sever Armenia’s ties with the Armenian Orthodox Church, a narrative cynically designed to weaponize religious identity and stoke fear that European alignment would threaten centuries of spiritual tradition. Perhaps the most corrosive was the institutional trust cluster. Much of this disinformation was explicitly designed not to persuade voters toward any particular outcome, but simply to erode confidence in the electoral process itself. The narratives aimed to undermine democratic practices and break trust in institutions. This form of disinformation is arguably the most durable, because even after a vote is concluded and results are certified, the doubt it plants can continue to linger.

What made the 2026 campaign distinct from previous Armenian information environments was not simply the volume of disinformation, but rather the sophistication of its delivery infrastructure. Politically affiliated networks created fake media websites, impersonated journalists and legitimate news outlets, and amplified false narratives through influencers and interconnected websites in an attempt to make fabricated stories seem credible. The result was a layered information environment in which it was genuinely difficult for ordinary users to differentiate between authentic journalism, domestic political messaging, and foreign-produced fabrication. It is important to note that foreign and domestic disinformation did not operate in isolation; they fed each other, amplified each other’s emotional registers, and collectively produced an information space that was structurally hostile to nuance.

Despite the scale of the operations, not all of the disinformation had its intended effect. The relationship between disinformation and public belief is not automatic, as context, lived experiences, and pre-existing trust levels all shape how different narratives land. International observers reported that authorities took steps to address disinformation, though the transparency and effectiveness of these efforts were limited. Civil society organizations, independent fact-checkers, and media literacy initiatives did attempt to counter false claims in real time. The Armenian government itself used the same social media platforms carrying disinformation to run voter education campaigns. The fact that this was necessary at all is itself significant, as the information environment had become so contested that the state felt compelled to compete within it rather than regulate from above.

Armenia’s trajectory with social media mirrors a pattern visible across post-Soviet and democratising states: early adoption as a tool of liberation, followed by instrumentalization as a tool of control, polarisation, and manipulation. The 2018 Velvet Revolution demonstrated the emancipatory ceiling, and the 2021 and 2026 elections demonstrated the adversarial floor. The most plausible near-term trajectory, hence, is not a resolution of this tension but the entrenchment of it. This is especially as short-form video content continues to grow in popularity, particularly among younger Armenians, meaning the speed at which emotional narratives can be distributed will outpace institutional fact-checking capacity. AI-generated multimedia lowers the production cost of disinformation to near-zero. And as long as the underlying condition persists, a population with high social media use but historically low institutional media trust, every new platform becomes a new vector for the same structural vulnerability. The hopeful reading is not naive optimism. Armenia has a growing civil society, a generation politically formed by the experience of 2018, and a population that has now, demonstrably, lived through a major coordinated disinformation campaign and retained enough critical capacity to assess it. The question for the coming years is whether media literacy, platform accountability, and institutional reform can develop quickly enough to match the pace of the threat, or whether the conversation will remain fatally one-directional.

Source: Santiago Ferbel-Azcarate is a Senior Research Assistant at LINKS Europe Foundation.