As Bangladesh moves toward its 13th general election, expected in February next year, concerns are mounting over the growing threat of AI-generated misinformation.
Technology experts and election stakeholders fear that artificial intelligence could be used to distort facts, influence voter sentiment, and undermine the credibility of the entire electoral process.
During a series of dialogues organised by the Election Commission (EC) with journalists, civil society representatives, academics, and women leaders, participants identified AI misuse as a key risk. Chief Election Commissioner (CEC) AMM Nasir Uddin also acknowledged the challenge, saying that steps are being taken to mitigate potential threats.
However, experts question whether the EC’s preparations are sufficient. Professor Md Abdur Razzak, chairman of the Department of Computer Science and Engineering at the University of Dhaka, said the commission is “far behind” in tackling AI-driven disinformation.
“It’s already late. The EC should have formed a functional team of academia, industry professionals, IT experts, and election specialists to design a policy and take timely action,” he said.
A global phenomenon
The use of AI in spreading disinformation is not new. The issue gained global attention during the 2016 US presidential election, when “fake news” was widely discussed. AI tools now allow anyone to create convincing fake videos, images, or statements that are nearly indistinguishable from real content — making detection extremely difficult.
Bangladesh has already experienced similar misuse. Ahead of the 2024 election, two deepfake videos falsely showing female candidates withdrawing from races in Bogura-2 and Gaibandha-1 went viral, misleading around 700,000 voters.
During the Dhaka University Central Students’ Union (Ducsu) polls, 51 pieces of misinformation were recorded — most of them false or manipulated.
Globally, the impact has been even more severe. Romania’s 2024 presidential election was cancelled due to AI-generated disinformation, while Australia, the US, and Canada also reported deepfake interference ahead of their national polls.
Rising digital reach
Bangladesh’s digital penetration amplifies the risks. A Bangladesh Bureau of Statistics survey shows that 99% of households now have mobile phones, 74.9% use smartphones, and nearly 60% have internet access.
The Election Commission’s data indicates that about 4 crore voters — roughly one-third of the electorate — are aged 18 to 33 and are highly active on social media. As election campaigning intensifies, AI-generated propaganda could spread faster than ever, shaping opinions before authorities can respond.
Surge in AI-based propaganda
Fact-checking organisation Rumour Scanner Bangladesh has identified 286 fake contents created using AI or deepfake technology in just the first nine months of this year. Among them, 77 were politically motivated.
“We’ve never seen this volume of politically driven fake content before,” said Tanvir Mahtab Abir, senior fact-checker at Rumour Scanner. “Detection is extremely difficult as many visual manipulation tools are sophisticated and tracing the original creators is nearly impossible.”
Weak EC preparedness
While the Election Commission has acknowledged the risk, its visible preparations remain limited. Officials say concrete actions will start in November, with training for election officers under the UNDP-IFES-supported “BALLOT Project.”
Election Commissioner Brig Gen (retd) Abul Fazal Md Sanaullah, who is leading the initiative, said recently awareness and counter-mechanism programmes would begin soon. “AI has become a global problem,” he said. “We plan to launch programs both centrally and locally from November.”
However, when this reporter contacted for two days to know the update of the programs, he did not respond.
Meanwhile, EC insiders admit that coordination gaps persist. The ICT Division’s System Manager Md. Rafiqul Haque, said he was unaware of any formal role assigned to his team regarding AI monitoring.
Senior Secretary Akhtar Ahmed said a workshop would be held on 22 October with experts from BUET, Dhaka University, MIST, and CID to seek advice.
Urgent call for action
Experts urge the EC to act swiftly. Professor Razzak said digital security must be reinforced — from protecting the NID database and EC website to formulating clear laws and communication strategies.
“You can’t build a system overnight,” he warned. “Once the campaign begins, it’ll be too late to control behaviour. Action must start now.”
