Bangladesh’s Proposed Cyber Law Pits Deepfake Protections Against Free Speech

Bangladesh’s proposed cyber law would expand penalties for AI-generated abuse, including material targeting women. The harder question is whether its broad rules on defamation and “rumors” could also revive the online speech restrictions that earlier laws were criticized for enabling. The draft is under review by a five-member cabinet committee, after media figures, legal analysts and civil society representatives urged officials to remove its most contested provisions.

The ICT Division published the draft Cyber Protection (Amendment) Act for public comment in mid-September. The committee is led by Posts, Telecommunications and ICT Minister Faqir Mahbub Anam. The government says tougher rules are needed to address harm from fabricated images, video and audio. Critics argue that vague language could expose journalists and other speakers to criminal charges.

Photo by cottonbro studio. Used under a Pexels License.

Deepfakes are already causing measurable harm

The case for addressing synthetic abuse is concrete. Farzana Sithi, a student activist associated with Bangladesh’s 2024 uprising, was targeted through a fake Facebook profile. A Dismislab investigation found that 133 of the 166 videos and reels posted to the profile between December 13, 2024, and March 17, 2025, were AI-generated deepfakes. The clips placed her face onto videos of two Indian content creators. The five most-viewed videos attracted more than 4.1 million views, and the account earned money through Facebook’s Stars feature.

“These videos are an attempt to tarnish my character,” Sithi told Dismislab. She said the material led to bullying and public shaming. She reported the profile to Facebook repeatedly, but it remained active at the time of the investigation. IT specialists she contacted privately asked for USD 100 to USD 150 to remove the fake accounts and videos, a cost she said she could not afford. Comments on the reels included rape threats.

Political women have faced related attacks. Before the February 12, 2026 election, a Daily Star analysis found AI-generated images targeting female National Citizen Party leaders with fabricated scenes implying moral or sexual impropriety. After the vote, a Prothom Alo fact-check identified fabricated images and statements aimed at candidates for reserved women’s seats. Rumor Scanner’s Tanvir Mahatab Abir told the outlet that women in public life are frequent targets of disinformation.

Synthetic media has also been used to attract social media followers for profit. Deepfake videos of Prime Minister Tarique Rahman and his family, including his daughter Zaima Rahman, appeared in fake “Family Card” offers on Facebook. Of 52 posts examined by the Daily Star, 13 featured Zaima. One page administrator told Dismislab the goal was to build an audience quickly and then sell the pages. That example shows how platform monetization can reward fabricated content even when it has no clear political purpose.

The draft raises penalties and widens the scope

Section 25 of the Cyber Protection Act, enacted on April 10, 2026, and amended on June 30, already covers blackmail, sexual harassment, revenge porn, child sexual abuse material, sextortion and the distribution of AI-generated or manipulated content. The minister has said those offenses are punishable by up to two years in prison, or up to five years when the victim is a woman or someone under 18. As written, the law addresses AI when it is used to commit specified offenses; it does not create a separate framework for synthetic media or political deepfakes.

The amendment would add defamation, humiliation and bullying to Section 25, while explicitly referring to AI-generated content. Prothom Alo has reported that the maximum penalty would increase to five years in prison and a BDT 2 million fine, roughly USD 16,131. If the target is a woman or a person under 18, the proposed maximum would rise to ten years and BDT 4 million.

A new Section 26A would address publishing or circulating a “rumor” or “disinformation.” It would allow a sentence of up to ten years, a fine of BDT 4 million, roughly USD 32,262, or both. The draft defines a rumor as unverified information that causes, or risks causing, confusion, panic or social instability.

In June, when introducing an earlier amendment, the ICT minister said platforms such as Meta would have to remove disputed material within a specified period. Officials have described the proposed penalties as a way to deter fake AI-generated video, audio and images. But the draft’s reach depends on how broadly offenses such as defamation and rumor are interpreted.

Past laws shape the debate

Bangladesh’s record of policing online expression is a central part of the response to the proposal. Section 57 of the Information and Communication Technology Act of 2006 penalized online content considered defamatory, offensive to religious sentiment or harmful to the state’s image. A 2013 amendment made the provision stricter, with sentences of up to 14 years. Journalists and ordinary social media users were among those charged.

After years of protest, the government promised to repeal Section 57. Yet when the Digital Security Act (DSA) took effect in October 2018, similar offenses appeared in Sections 25, 28, 29 and 31. By 2023, the law minister had told parliament that 7,001 cases had been filed under the DSA. Writer Mushtaq Ahmed died in custody in 2021 after spending ten months in jail under the law, prompting protests in Bangladesh and abroad.

The Awami League government replaced the DSA with the Cyber Security Act in 2023. Amnesty International said the new law retained almost all of the DSA’s repressive features. Following the July 2024 uprising, the interim government used a 2025 Cyber Security Ordinance to repeal nine sections. Rights groups say hundreds of cases from earlier years remain pending.

Against that history, Transparency International Bangladesh has warned that terms such as “rumor” and “demeaning” could permit selective enforcement. Rights organizations have said the draft echoes the repealed DSA. The Committee to Protect Journalists has objected to criminal liability for reporting that officials might consider unverified. An Asia Times analysis also pointed out that because defamation would sit within Section 25, the higher penalties for offenses against women could apply to defamation as well as sexual abuse.

Scale of abuse, gaps in support

The need for effective remedies is substantial, though available figures don’t isolate AI-related cases. The Dhaka Tribune reported that police’s Cyber Support for Women unit had received approaches from 60,808 women by September 2024. A 2024 survey by the Bangladesh Bureau of Statistics and UNFPA found that nearly 89 percent of women who use social media had experienced online violence at least once. Officials say underreporting means the actual scale is probably larger.

A regional study published in the Journal of Online Trust and Safety drew on three workshops with 49 journalists, lawyers, fact-checkers and advocates in Bangladesh, India and Sri Lanka. It found that women are disproportionately targeted, platform moderation often performs poorly in South Asian languages, and victims can struggle to obtain institutional help. Sithi’s experience illustrates the gap between reporting abuse and securing a remedy when paid technical assistance is out of reach.

Access to digital skills and internet service also varies. The Asian Development Bank reports functional digital literacy of 49 percent in Bangladesh and says rural residents are 29 percent less likely than urban residents to use mobile internet. Bangladesh Bureau of Statistics figures released in April put internet use at 56.6 percent among men and 50.2 percent among women. Advocates say limited digital skills can make it harder to recognize manipulated media or seek help, though these surveys do not measure exposure to deepfakes.

Fact-checkers are trying to respond, but their capacity is constrained. FactFest 2026, held on September 27 and 28, brought together around 100 fact-checkers, journalists and researchers. Its 13 sessions included AI and gendered disinformation. At the closing roundtable, participants described shrinking funding and limited coordination with mainstream newsrooms.

The decision ahead

The cabinet committee has not finalized the draft. If approved, it would proceed to Parliament. Constitution Watchdog has urged the government to withdraw the current version and reopen public consultation for at least 30 days. The group acknowledged the stated aim of protecting women and children from online abuse and responding to deepfakes, while calling for changes to the proposal.

Bangladesh’s challenge is to make reporting and removing abusive synthetic media practical for victims, while ensuring that ordinary criticism and journalism don’t become grounds for prosecution. The committee’s revisions will show whether the government can address a real and growing harm without giving officials broad discretion over online speech.