The 13th parliamentary election, held on 12 February, was marked by a sophisticated and intensifying wave of misinformation that evolved from questioning the feasibility of the polls to contesting the legitimacy of the results.
A comprehensive analysis by nine fact-checking organisations documented 1,185 unique pieces of political misinformation between 11 December 2025 and 20 February 2026, compiled by Dismislab.
Nearly half of these cases were directly tied to the election, with approximately 12% involving content generated or manipulated by artificial intelligence (AI).
Trajectory of falsehoods
The volume of misinformation fluctuated in alignment with the electoral calendar. In the early phases, from the announcement of the schedule on 11 December through early January, false claims remained relatively moderate, averaging 10 to 12 cases per day.
However, as the polls drew closer, the frequency of election-specific misinformation accelerated sharply.
The formal campaign period accounted for the largest overall share of election-related misinformation at 41%.
Yet, the most concentrated surge occurred during the legally mandated 48-hour silence period immediately preceding the vote. During these two days, 95 election-related false claims were documented, representing nearly 18% of the entire election-specific dataset.
On 11 February alone, over 70 combined political and election-specific claims were debunked, marking the highest daily spike of the cycle – a six to seven-fold increase over the early January baseline.
The nature of the misinformation shifted as the process moved through different milestones.
Pre-election and nomination
Early rumours focused on whether the election would take place, with claims of postponement or cancellation. One AI-generated image falsely depicted a torch procession by the Shariatpur District Awami League demanding a schedule cancellation.
During the nomination phase, focus shifted to candidate eligibility. Claims circulated that the nominations of Bangladesh Jamaat-e-Islami Ameer Shafiqur Rahman and BNP Chairman Tarique Rahman (now Prime Minister) had been cancelled.
False reports also alleged that Jamaat Secretary General Mia Golam Parwar intended to introduce Pakistan’s legal system in Bangladesh.
Misinformation intensified to include fabricated candidate promises. These included claims that Tarique Rahman would close the metro rail or that free data packages would be distributed.
Religious framing also emerged, with forged cards suggesting the Jamaat chief had threatened to send Hindus to India if they did not vote for the party. Additionally, NCP leader and Dhaka-11 candidate Md Nahid Islam faced false allegations regarding dual citizenship in Dominica.
Silence period and election day
Narratives turned toward immediate disruption and fearmongering, using old footage to represent unfolding violence. On Election Day, false reports claimed polling stations were being seized or that candidates, such as Nurul Islam (BNP, Barguna-2), had withdrawn.
AI-generated videos were specifically used to fabricate withdrawal announcements for independent candidates Ataur Rahman Khan Angur (Narayanganj-2) and Samiran Dewan (Khagrachhari).
Post-Election Contestation Following the vote, the narrative shifted from “will it happen?” to “was it stolen?”. Misinformation focused on vote legitimacy and alleged manipulation.
Social media posts claimed Jamaat-e-Islami had been defeated through rigging in 53 seats where the margin was supposedly under 5,000 votes.
However, an analysis by Dismislab found that only 22 constituencies were determined by such a margin, with 11 going to Jamaat and its allies and nine to the BNP and its allies.
Further false claims involved nationwide recounts in 150 constituencies and fabricated reports of post-election violence against Hindu communities. Even cabinet formation was targeted with false reports assigning ministerial portfolios to various figures.
Tactics and AI integration
Misinformation was frequently packaged to mimic professional journalism. Approximately 54.5% of documented cases used graphic cards designed to resemble legitimate news media layouts, while 30.3% were video-based.
AI played a steady role throughout the cycle, accounting for 12% of election-related cases. Nearly 70% of AI-linked misinformation was video-based, often styled as news broadcasts or security footage.
While the silence period saw the highest overall spike in claims, it was largely text-heavy, likely because procedural rumours are faster to distribute than synthetic media.
This report is based on findings from nine verification organisations – Rumor Scanner, BoomBD, Newschecker, Fact Crescendo, FactWatch, AFP Bangladesh, Ajker Patrika, Dismislab, and The Dissent. The data covers the period from 11 December 2025 to 20 February 2026.
