In an era where technology and politics are becoming increasingly intertwined, Bangladesh is witnessing the rise of a troubling phenomenon: AI-generated political videos that blur the line between truth and fiction.
As the country heads toward its next general election next year, experts warn that AI technology could become a powerful tool for misinformation, disinformation, and digital manipulation in the hands of political actors.
They suggested that, before the election, the government or the Election Commission should establish guidelines on the use of AI in political parties’ election campaigns; otherwise, there is a risk of chaos.
In recent months, several suspiciously edited videos have circulated on social media, purportedly showing politicians making outrageous remarks or engaging in controversial actions.
While journalists and fact-checkers quickly debunk many of them, the initial damage—public confusion, outrage, or polarization—often lingers.
According to an independent online verification and research platform, Dismislab.com, since mid-June 2025, Bangladesh has witnessed a surge in AI-generated political campaign videos—synthetic clips featuring entirely fictional people urging voters to support specific parties, particularly Jamaat-e-Islami.
These highly realistic videos simulate real interviews with “ordinary citizens” from diverse backgrounds—Hindu women, rickshaw pullers, fruit sellers, and others—presenting a fabricated image of broad public support for Jamaat and, later, other parties like BNP, NCP, Islami Andolon, and even AL supporters.
Dismislab analyzed 70 such videos, which collectively garnered over 23 million views.
Initially led by Jamaat supporters aiming to rebrand the party across social and religious lines, the trend quickly expanded.
BNP, NCP, and even AL-aligned users began producing their own synthetic content—some promoting candidates, others attacking their rivals.
The content included identity-based appeals, religious messaging, satirical takedowns, and false allegations—none of which were supported by evidence.
Beyond political parties’ campaigns, some AI-generated content has misrepresented Chief Adviser Professor Muhammad Yunus, BNP Acting Chairperson Tarique Rahman, Awami League President Sheikh Hasina, and other top Bangladeshi politicians.
One such video, allegedly showing Dr Yunus, appears to invite people to gamble for financial gain.
Later, the Chief Adviser’s Press Wing urged social media users to be cautious and avoid being misled by deepfake videos depicting Professor Muhammad Yunus.
“Gambling ads targeting Bangladeshi users on social media aren't new, but they’ve taken a disturbing turn—now exploiting the huge popularity of Professor Muhammad Yunus, the Chief Adviser of the Bangladesh government. Gamblers have been producing AI-generated deepfake videos to mimic news reports and fake statements in a bid to attract the public to their sites,” the press wing said in a statement debunking the video.
However, creating such content no longer requires technical expertise. With the advent of free and low-cost AI tools like HeyGen, DeepFaceLab, and Synthesia, anyone with a smartphone and internet connection can now produce videos that convincingly mimic the appearance and voice of real people.
Despite the potential impact, Bangladesh currently lacks a legal or regulatory framework specifically targeting AI-generated misinformation.
The Digital Security Act—controversial for its broad language and alleged misuse against journalists and dissenters—does not directly address deepfakes or AI-generated videos.
The Bangladesh Election Commission, which has long struggled with issues surrounding campaign financing, posters, and election codes of conduct, has yet to issue specific guidelines regarding the use of AI content in political campaigns.
When contacted, Qadaruddin Shishir, one of Bangladesh’s leading fact-checkers, suggested that the government or Election Commission should introduce a legal framework addressing the use of AI videos before the election.
“Positive election campaign videos made using AI can be allowed if they include a specific disclaimer, but negative or rivalry campaigns or hate speech videos should be banned through a legal framework,” Shishir added.
Bangladesh is not alone. In India, deepfakes of political leaders went viral during the 2024 general elections, prompting calls for stricter AI governance.
In the US and Europe, tech companies are already being pushed to watermark AI-generated content and flag manipulated media ahead of national elections.
AI video campaigning in elections has become a rising concern globally—not just in Bangladesh.
In the United States, 25 states have introduced laws to regulate the use of AI in elections.
Texas and Minnesota specifically ban the release of political deepfakes within a specific timeframe before voting.
Meanwhile, 23 other states mandate that media content must include a disclosure if it contains AI-generated elements, similar to existing requirements for political advertising.
Before the 2024 European Parliament elections, the European Union’s AI Act also addressed AI-related threats to democracy.
The Act categorizes AI systems into four risk levels, banning manipulative or exploitative uses and designating AI tools that can directly influence voters’ choices as “high-risk” systems.
But Bangladesh’s case is more delicate. With millions of first-time internet users, low levels of digital literacy, and a highly politicized media environment, the country is fertile ground for disinformation.
“Digital literacy is around 10 percent in Bangladesh, so this is a growing concern,” Shishir further added.
When contacted, Fahmidul Huq, a faculty member at Bard College, USA, told Dhaka Tribune that the use of AI videos is turning into a pandemic of disinformation.
“The challenge of fact-checking is growing. We are also seeing this in the political sphere. Before elections, a large amount of content will be created on digital media—this is natural.”
Fahmidul, a former professor at Dhaka University, mentioned the ethical challenges of AI-generated videos and said: “Presenting nearly 'impossible' scenarios in a normal way will fuel ethical debates—for example, making a person speak in favor of a party or its symbol, which is almost impossible in reality. In previous elections in various countries, concerns were raised about deepfakes—making someone say something they never actually said or portraying them in situations they were never in.”
“Objections were raised, and in some cases, bans were imposed. Creating deepfakes used to be expensive and time-consuming. But AI is now making the same task much easier. There is ample reason for concern, especially since misleading the general public will become very easy.”
He also suggested that the Bangladesh Election Commission must establish guidelines on this matter.
“For instance, such promotional content must be labeled as 'AI-generated' or 'created using AI technology.' On the other hand, political parties should issue clear statements or commitments on this issue.”
When contacted, Shafiqul Alam, press secretary to the chief adviser, told Dhaka Tribune that guidelines for AI-generated videos are essential, as a significant number of such videos have recently appeared on social media, spreading misinformation and disinformation.
“Hopefully, the Election Commission will take the AI video campaign issue seriously and form a guideline after consulting with local and international experts,” he added.