On March 5, 2026, Nepal held its first general election since the youth-led "Gen Z" uprising of September 2025 — protests that began over a brief social media ban, left dozens dead, and ultimately toppled the government of then-Prime Minister KP Sharma Oli. An interim government led by former Chief Justice Sushila Karki was formed to steer the country to this snap election. It should have been a moment for Nepal's democracy to show it could recover and renew itself. Instead, the campaign period became something else entirely: one of the clearest real-world demonstrations yet of how AI-generated deepfakes and disinformation can flood a national election in a country with limited fact-checking infrastructure.
This is a look back at what actually happened — how much false content circulated, the specific fabricated incidents that went viral, why Nepal was especially exposed, how authorities and independent fact-checkers responded, and what it means for how the next election cycle should be handled differently.
The Scale of the Problem — How Much Misinformation Circulated
The numbers tell a clear story of escalation. In late January 2026, roughly a month before polling day, the Election Commission of Nepal's Information Integrity Promotion Centre had already identified over 300 cases of misinformation, disinformation, and hate speech circulating online, with 21 parties or individuals asked to clarify possible code-of-conduct violations. By the time voting actually took place on March 5, that number had roughly doubled — election officials reported that more than 600 cases had been referred to authorities, with around 150 of those handed specifically to the police for investigation.
Independent fact-checking organisation NepalFactCheck.org, which has operated since 2020, described a multifold increase in misinformation volume compared to previous election cycles. In response, the organisation scaled up significantly, publishing 60 fact-checks in the 30 days leading up to voting day alone, plus 8 more on election day itself — an intensity of work it hadn't needed in prior elections. Media analysts tracking the trend on national television described the sheer numerical increase in false content as unlike anything seen in previous Nepali elections.
Notable Incidents: Fabricated Images and Videos Targeting Candidates
Several specific fabrications stood out for how far they spread before being debunked:
- A manipulated photo of a Rastriya Swatantra Party candidate from Chitwan, falsely labeled as showing her as a foreign "agent." Fact-checkers traced the real photo back to a public youth policy dialogue event from mid-2025, co-organised with a foreign embassy's youth council, where she had simply been an invited speaker.
- A fabricated video appearing to show three of Nepal's most recognisable political figures meeting together, fueling speculation about a new political alliance. The meeting depicted in the AI-generated clip never actually took place.
- A deepfake audio clip fabricating a private conversation between a senior party leader and a foreign head of government, designed to imply an inappropriate or controversial relationship that didn't exist.
- AI-generated "aerial" images inflating rally crowd sizes — one widely shared image claimed to show a gathering of roughly 500,000 supporters for a former prime minister's party; the actual crowd was found to be far smaller.
- A manipulated video circulated on TikTok appeared to show one opposition leader endorsing a rival party, a claim that was also later debunked.
- Manipulated images claiming a senior opposition leader was publicly "chased away" by an angry crowd were confirmed by fact-checkers to have been artificially generated rather than real footage.
Beyond individual candidates, fact-checkers and researchers also flagged influence campaigns pushing narratives from outside Nepal entirely, including content amplifying calls to restore the country's former Hindu monarchy — a reminder that election disinformation wasn't limited to domestic party rivalries alone.
Why Nepal Was Especially Vulnerable
Several factors combined to make Nepal an especially exposed environment for AI-driven disinformation. Roughly 56% of the population is online, and an estimated 80% of that internet traffic flows through just a few platforms — Facebook, Instagram, and TikTok — meaning a single viral post can reach an enormous share of the electorate within hours. At the same time, digital literacy has not kept pace with connectivity: earlier national survey data found that over 95% of online users had encountered false information at some point, yet fact-checking and media literacy infrastructure remained comparatively thin going into this election.
Nepal's situation was also worsened by inconsistent support from major social media platforms. Compared to the 2022 general election — when one major platform had opened its ad library, trained officials and journalists, and actively supported misinformation monitoring — platform engagement during the 2026 cycle was markedly thinner, limited mostly to a brief visit from one platform and a virtual meeting between another and the Election Commission. Fact-checkers publicly noted that Nepal, like many countries outside major Western markets, often receives limited proactive support from global platforms during exactly the moments — elections — when that support matters most.
Compounding this, fact-checkers noted that AI detection and deepfake-verification tools remain far less mature for Nepali-language content than for English or Hindi, making some fabricated audio and video harder to conclusively verify or debunk quickly.
What Authorities and Fact-Checkers Did in Response
The Election Commission of Nepal set up a dedicated monitoring body, the Information Integrity Promotion Centre, to track digital violations of the Election Code of Conduct and request the takedown of suspicious content from major platforms. Officials confirmed that hundreds of cases were referred onward, some resulting in fines, formal clarifications requested from parties, and at least one detention of an individual accused of using social media posts to intimidate potential voters. Even so, election officials and outside observers acknowledged that the sheer volume of disinformation outstripped what fines and case-by-case enforcement could realistically contain.
Independent fact-checkers played an outsized role given the limits of official enforcement. NepalFactCheck.org secured emergency funding, including support from the International Fact-Checking Network, to hire additional fact-checkers and maintain nationwide monitoring through the campaign period. It also formalised a publishing partnership with a leading Nepali news portal, while one of the country's largest media houses launched its own dedicated fact-checking section around the same time. Fact-checkers used a mix of manual visual inspection — looking for unnatural hands, faces, or lighting — combined with AI-detection tools to confirm suspected fabrications before publishing corrections. Separately, international development partners supported the Election Commission with AI-powered social media monitoring tools to help identify emerging disinformation trends, hate speech, and code-of-conduct violations in near real time.
Despite the scale of the problem, the election itself proceeded and was completed peacefully, and independent fact-checking reviews afterward found no conclusive evidence that the disinformation wave had directly changed the final election outcome.
Lessons for the Next Election Cycle
International election observers who assessed Nepal's 2026 process concluded that the digital information environment was the single fastest-moving integrity risk of the entire election — evolving faster than the country's institutions could realistically monitor or respond to. Several lessons stand out for future cycles. First, platform engagement needs to be secured well in advance and treated as non-negotiable, rather than left to ad hoc, last-minute outreach close to polling day. Second, independent fact-checking capacity needs sustained, advance funding rather than emergency fundraising once a crisis is already underway — Nepal's fact-checkers only scaled up because of last-minute grants and volunteer effort. Third, AI-detection tools built for Nepali-language audio and video need dedicated investment, since fact-checkers explicitly flagged that existing tools work far better for English and Hindi content.
Finally, observers cautioned that any strengthened monitoring and enforcement must remain transparent, rights-consistent, and civilian-led — the concern being that heavy-handed content policing, if implemented carelessly, could itself become a tool for suppressing legitimate political speech rather than protecting it. Getting that balance right, between containing genuinely fabricated content and preserving open political debate, is likely to be the central challenge for Nepal's electoral institutions heading into future elections.
Frequently Asked Questions
Did AI deepfakes change the outcome of Nepal's 2026 election?
Independent fact-checking reviews conducted after the election found no conclusive evidence that the disinformation wave directly altered the final results, and the vote itself proceeded peacefully despite the scale of false content circulating during the campaign.
How many misinformation cases were reported during the election?
The Election Commission of Nepal reported that more than 600 cases of misinformation, disinformation, and hate speech were referred to authorities by election day, with around 150 of those forwarded specifically to police for investigation.
What kinds of deepfakes were used against candidates?
Documented cases included fabricated or mislabeled candidate photos, an AI-generated video depicting a political meeting that never occurred, a deepfake audio clip fabricating a private conversation, and artificially inflated crowd-size images shared to exaggerate rally attendance.
Why couldn't platforms like Facebook or TikTok stop the disinformation?
Fact-checkers noted that major platforms provided noticeably less proactive support during this election compared to previous cycles, with engagement limited mostly to brief visits or virtual meetings rather than the more active monitoring partnerships seen in earlier years.
Who fact-checked the election in Nepal?
NepalFactCheck.org, an independent fact-checking initiative operating since 2020, led much of the verification effort, publishing dozens of fact-checks in the month before voting with support from international fact-checking funding networks and partnerships with local media outlets.
Discussion