March 2, 2026
DHAKA – Bangladesh’s 13th National Parliamentary Election was conducted within a “digitally saturated” environment where communal misinformation and AI-generated deception were weaponised to manipulate voter perception, said Activate Rights, in a report released yesterday.
Activate Rights is a non-profit initiative on protecting digital rights in Bangladesh.
The study, titled “Faith, Fear, and Falsehoods: Mapping Communal Misinformation and Hate in Bangladesh’s 2026 Election”, analysed data from December 11, 2025, to February 16, 2026 and concluded that coordinated narrative manipulation functioned as a structural feature of the February 12 election.
The polls saw a disturbing rise in the use of generative AI to target female candidates, found the report. “Women candidates experienced disproportionate levels of online abuse despite constituting only four percent of contestants, with seven ultimately elected,” said the report.
Dhaka-9 candidate Tasnim Jara and Brahmanbaria-2 candidate Rumeen Farhana, who won the seat, were among the victims of AI-generated explicit images.
AI was also used to spread misinformation. AI-generated minority characters were shown claiming they are in danger, or asserting that if people don’t vote for Jamaat, they will be sent to India, found the report.
The report stated that alongside communal misinformation, election-time discourse from hardline Islamist accounts on social media sought to delegitimise democratic participation by claiming democracy and voting are fundamentally incompatible with Islam and thus religiously forbidden.
The report also reveals that the Awami League’s exclusion from the election, following ban on its political activities, prompted a coordinated social media boycott using the hashtag #NoBoatNoVote. But it went beyond a symbolic boycott, moving toward explicit obstruction. For instance, a post directly encouraged supporters to resist the vote, and a TikTok video featured lyrics that promoted physical aggression against ballot boxes.
In addition, election discourse featured systematic hate speech against political parties and candidates. Jamaat-e-Islami and allies were labelled “extremists” with derogatory content circulating after a party leader’s social media account was reportedly hacked. BNP figures were widely called “extortionists”, with posts claiming “70 percent” were criminals.
Another tactic involved circulating selectively edited video clips of political figures.
These non-deepfake excerpts removed context, leading to captions accusing speakers of blasphemy or anti-Islamic sentiment. Comments frequently used labels such as “atheist”, “kafir”, or “apostate”, showing how contextual distortion quickly generates hate-driven engagement.
The report highlights a significant cross-border dynamic, noting that many narratives originating in India portrayed Bangladesh as a hotbed of extremism and systemic minority persecution.
These narratives often utilised recycled or miscontextualised footage — such as reframing a food stall dispute in Chattogram as an attack on Muslim girls for not wearing hijab, which was later found not related to religion and hijab.
Not only on Facebook, X accounts in India claimed “Islamists” were forcibly cutting students’ hair for wearing western outfits. The footage actually depicted a symbolic protest in Mymensingh against a specific individual’s haircut and had no religious or extremist motivation.