AI-generated videos spread disinformation online, authorities yet to act

A middle-aged woman with one arm missing appears in a video posted on a Facebook page called Uttarbanga Television. In the video, she claims that leaders of a political party took money from her, promising to arrange a disability allowance card, but never delivered it. She further says she will no longer vote for that party and explicitly names the party she intends to vote for instead.

The video is not authentic. Prothom Alo, after fact-checking, has found that it was created using artificial intelligence (AI). However, the woman’s appearance closely resembles that of Rikta, a garment worker who lost her arm in the Rana Plaza collapse in Savar in 2013.

The video was uploaded to the Uttarbanga Television Facebook page on 10 January. As of around 5:45 pm yesterday (Sunday), it had been viewed nearly one million times. About 55,000 people reacted to the video, more than 1,200 comments were posted, and it was shared over 21,000 times.

According to a report by fact-checking organisation Dismislab, the Uttarbanga Television page has more than 90,000 followers (which rose to 106,000 by yesterday evening). The page is run by seven ‘admins, all operating from Bangladesh.

The Representation of the People Order states that in this election, creating, publishing, promoting, or sharing false, misleading, biased, hateful, obscene, offensive, or defamatory content using artificial intelligence to mislead voters is prohibited. This restriction applies not only to political parties and candidates but also to any individual acting on their behalf.

Despite this, such prohibited campaigning is continuing unhindered. In some cases, AI-generated videos show people dressed as military officers urging voters to support a particular party.

In others, women in bikinis claim they will vote for an Islamist party—content that is clearly damaging to that party’s image. AI-generated videos featuring religious minorities are being used to criticise one party and promote another. In yet another instance, a monkey is shown urging votes for a political party.

There is no data on how many such videos have spread, how many people have watched them, or what percentage of viewers believe them, as no research has been conducted. However, the videos have garnered thousands of views, indicating a large audience.

According to the Bangladesh Telecommunication Regulatory Commission (BTRC), Bangladesh has approximately 129.9 million internet users (anyone who has used the internet at least once in the past 90 days is counted).

Data from the website Hashmeta suggests that Bangladesh may have over 45 million Facebook users, between 15 and 18 million TikTok users, and around 35 million YouTube users.

In a televised address to the nation on 12 December, Chief Election Commissioner (CEC) A M M Nasir Uddin spoke about the spread of false information using AI. He said misinformation and disinformation created through artificial intelligence are increasing day by day and that necessary action would be taken under the law to prevent this.

However, no action against anyone has been reported so far. When contacted about the matter, the Chief Adviser’s press wing said responsibility for the issue has been assigned to the Ministry of Information and Communication Technology.

Faiz Ahmad Taiyeb, special assistant to the Chief Adviser in charge of the ministry, told Prothom Alo yesterday evening, “We have launched four email addresses and one hotline number to receive complaints from citizens. We are processing the reports we are receiving.”

Voting for the 13^th National Parliament election will take place on 12 February. Official campaigning begins on 22 January. Before that, asking for votes for any electoral symbol is not permitted, and symbols have not yet been allocated. Despite this, campaigning is underway on social media platforms such as Facebook, TikTok and YouTube, with calls to vote for specific symbols.

Asked about the issue, election expert and former member of the Election System Reform Commission, Md Abdul Alim, told Prothom Alo that the law does not clearly state whether online campaigning is allowed before symbol allocation.

He said the Election Commission can take action if such situations arise, but no such action is currently visible. He added that using AI for dishonest purposes is prohibited and that steps must be taken before the situation worsens.

Asked how the situation could deteriorate further, Abdul Alim said, “Suppose an AI-generated video is spread to humiliate a candidate—his supporters will not tolerate it. That could lead to violence. Or imagine an AI video claiming a candidate has withdrawn from the election—what would happen then?”

AI Videos Showing Military Uniforms Used in Campaigning

One widely circulated video shows a man wearing the uniform of the Bangladesh Army urging support for a particular party. Fact-checking has confirmed the video is AI-generated. As of yesterday, it had been viewed more than one million times, received over 89,000 reactions, attracted about 2,500 comments, and been shared 26,000 times.

On 14 January, the Bangladesh Army issued a warning about such AI-generated videos. In a statement posted on its official Facebook page, the army said that vested quarters are spreading AI-generated content and misleading photo cards on social media to damage the army’s image, and urged the public not to be misled.

Four days have passed since the army’s warning, yet the video remains online.

Political parties say they have raised the issue of AI misuse with the Election Commission, but no effective action has followed. Jamaat-e-Islami assistant secretary general and head of its publicity and media department, Ehsanul Mahbub Jubair, told Prothom Alo that social media platforms are open spaces, making it difficult to identify who is doing what. He said, “Jamaat is being targeted the most through attacks and disinformation using AI. We have raised the issue multiple times in meetings with the Election Commission, but the steps taken are not satisfactory.”

‘No Option to Ignore This New Reality’

An AI-generated video featuring National Citizen Party (NCP) leaders and parliamentary candidates Hasnat Abdullah and Sarjis Alam is also circulating, showing the two dancing shirtless in lungis. Fact-checking confirmed that this video was also AI-generated.

A study on electoral systems in Southeast Asia, published in October 2025 in the journal Frontiers in Political Science, warned that AI-generated videos pose a major threat to free and fair elections. Citing the study, a Dismislab report said such videos are being used to spread fabricated stories about political leaders, which voters often accept as true. In elections in the Philippines and Thailand, AI-driven “bots” were used systematically to spread false information against opponents.

Similar concerns had already been raised in Bangladesh. On 23 December, during a meeting between the Election Commission and returning officers and field administration officials, challenges related to misinformation, rumours, and the misuse of social media and technology were discussed. However, no effective measures or visible initiatives have followed.

Professor Shamim Reza of the Department of Mass Communication and Journalism at Dhaka University told Prothom Alo that personal attacks, false claims, misinformation and disinformation against opponents have been seen in past elections as well.

However, due to the rapid expansion of information technology, the speed and scale of such campaigns have multiplied many times over. “There is no option to ignore this new AI-driven reality,” he said.

He added that young people who are active on social media—and who tend to rely more on social platforms than on traditional media analysis—may be more strongly influenced by such videos. Overall, he warned, there may be forces actively working in a planned manner to undermine or challenge the entire electoral framework.