A woman posts an ordinary photograph of herself on social media. Hours later, she finds that someone has run it through an AI tool and turned it into an explicit sexual image. The fabricated image is already circulating online. Strangers leave obscene comments. Others harass her. Someone is even demanding money to prevent its further distribution.
The image is fake. The humiliation, fear and psychological trauma are not.
Not long ago, faking a photograph took skill, time and specialised software. Today, tools available to anyone can produce a convincing fake image or video in minutes. In many cases, victims have no idea that their photographs are being used.
Deepfakes use AI to superimpose a person's face, voice or physical appearance onto another image or video, creating fabricated content that can appear authentic.
Although the technology has legitimate applications in entertainment and creative industries, its misuse for sexual harassment, blackmail and character assassination is also raising serious concerns.
The danger goes beyond manipulated images. It has become new ways to exercise control over women's bodies, identities, privacy and reputation.
A 2025 study by UN Women found that generative AI is increasing the scale, speed and severity of technology-facilitated violence against women, from deepfake pornography and automated hate speech to impersonation, doxxing and online blackmail.
The most disturbing change is that a perpetrator no longer needs access to a woman’s private images or videos. A public photo is enough. A woman who has never shared anything intimate can still be sexually violated online.
The consequences can be devastating. The victim often has to prove that the material is fake, protect her reputation and explain herself to family, colleagues and strangers. The person who made it stays anonymous. The burden of the crime lands on the one it was committed against.
SILENCING, BY DESIGN
The harm goes beyond the individual. Women who live public lives -- particularly journalists, human rights defenders, writers, and politicians -- are the easiest targets, and the fallout shows in the numbers.
A 2026 report titled Tipping Point: Online Violence Impacts, Manifestations and Redress in the AI Age by UN Women, UNESCO and the International Center for Journalists covering research across 119 countries in 2025, found that 5 per cent of surveyed women journalists had faced attacks involving deepfakes or manipulated images and videos.
About 45 per cent self-censored on social media because of online harassment, and 22 per cent did so at work.
The psychological consequences are equally serious. About 25 per cent of surveyed women journalists reported being diagnosed with or treated for anxiety or depression linked to online violence. Another 13 per cent reported experiencing post-traumatic stress disorder (PTSD).
These figures point to a chilling consequence: digital violence can push women out of public conversations.
A journalist may hesitate to express an opinion on social or political issue for fear that her photograph will be turned into sexually explicit content. A politician facing fabricated videos and sexual threats may find public engagement increasingly difficult.
This is how AI-enabled abuse silences women: it does not have to defeat their views, only intimidate them into silence, eroding their freedom of expression and their right to participate equally in society.
BANGLADESH'S GROWING CHALLENGE
Bangladesh is not immune to this emerging threat.
As women's participation in digital spaces grows, reports of manipulated images, identity impersonation, harassment and AI-assisted blackmail are becoming more common.
A 2026 study, "Victims of AI: Addressing the Legal Response Gap to AI-Related Crimes in Bangladesh", surveyed 435 people aged 18 to 35 and examined two cases of AI-generated deepfake harassment and blackmail.
It found that 55 per cent of respondents considered AI misuse very common, while 65 per cent believed existing laws and regulations were inadequate to address emerging AI-related harms.
Bangladesh has the law reaching this conduct.
Section 25 of the Cyber Protection Act, 2026, covers the transmission, publication or circulation of harmful or threatening material, including AI-generated and edited images, when used for blackmail, sexual harassment, revenge pornography, digital child sexual abuse material or sextortion.
Where the offence is committed against a woman or a child under 18, the proposed maximum punishment would increase from five years to 10 years' imprisonment, or a fine of up to Tk 40 lakh, or both.
However, tougher penalties alone cannot ensure protection.
Victims must be able to report offences quickly. Investigators need the expertise to identify perpetrators and preserve digital evidence. Police, prosecutors and courts must be equipped to deal with crimes involving deepfakes and AI-generated content.
Original files, timestamps, account information and records showing how fabricated images or videos were circulated can be crucial to investigations. Victims must also receive clear guidance on preserving such evidence and seeking assistance.
The proposed amendment also seeks to establish time limits and procedures for removing or blocking harmful online content. It proposes replacing the existing requirement for Cyber Tribunal approval after content is blocked under Section 8 with a review mechanism involving the relevant agency. The agency would have to provide reasons for blocking within three days and decide on a review request within 15 days.
Such safeguards matter because removing a fabricated image from one platform does not mean it has disappeared from the internet. It may already have been downloaded, shared through messaging groups or uploaded elsewhere. Effective and timely removal mechanisms must therefore complement criminal investigations and prosecution.
Bangladesh is also preparing a National AI Policy. According to the official National AI Polilcy website, the ICT Division formed a 16-member committee on November 2, 2026, to prepare a draft by December 15.
Preventing the creation and circulation of non-consensual sexual images, protecting women's digital rights and establishing accountability for AI misuse should be part to that policy.
TECHNOLOGY COMPANIES CANNOT LOOK AWAY
The responsibility does not rest with governments alone. Technology companies must also be held accountable for how their platforms respond to AI-enabled abuse.
Social media algorithms often amplify content that attracts engagement. Consequently, a fabricated sexual image can spread rapidly as users view, comment on and share it. The perpetrator may take minutes to create the image, while the victim may spend days trying to have it removed.
Platforms need effective mechanisms to detect AI-generated sexual content, respond promptly to complaints and remove abusive material upon verification. They should also cooperate with law enforcement agencies, within the bounds of the law, to help identify accounts and trace the sources of abusive content.
At the same time, efforts to tackle online abuse must not become a pretext for restricting legitimate expression, journalism or political criticism. Clear definitions, transparent complaint procedures and independent oversight are essential to protecting both women's safety and freedom of expression.
CONSENT CANNOT BE AN AFTERTHOUGHT
As AI technology advances, manipulating images and videos will become even easier, while distinguishing fabricated content from authentic material may become increasingly difficult. Technology alone cannot solve a problem that is also rooted in social attitudes and the abuse of power.
Educational institutions, media organisations and civil society groups must raise awareness about digital safety and consent. Young people, in particular, need to understand that creating, storing or forwarding AI-generated sexual images can also contribute to abuse.
Equally important is ending the tendency to blame victims. A woman's clothing, social media activity or decision to share an ordinary photograph does not give anyone the right to manipulate it for sexual exploitation. Public availability is not consent.
Governments, technology companies, law enforcement agencies and civil society must work together to ensure that victims can report abuse without fear, evidence is preserved, perpetrators are investigated and harmful content is removed.
AI was created to expand human possibilities. When it is used to attack someone's body, identity and dignity without consent, the issue is no longer merely technological. It is a question of rights, safety and justice.
-Md Abbas is freelance journalist and corporate communication professional.
Related Posts
Women in politics / Needed in protests, not in power
The July Uprising of 2024 brought thousands of women onto the streets of Bangladesh. Female students, teachers, professionals and ordinary citizens joined protests, organised marches and helped sustain a movement that eventually toppled the government. Many took considerable risks, facing arrest, harassment and threats to their safety.
- Opinion
What’s in the name?
Whenever governments change in Bangladesh, the names of state institutions and public infrastructure change.
- Opinion
Journalist killing / The state cannot escape responsibility for impunity
When journalists Sagar Sarowar and Meherun Runi were murdered in their Dhaka apartment on February 11, 2012, their five-year-old son Megh was at home. More than 14 years later, the investigation into their deaths remains unfinished. Megh is now 19, but the state has yet to establish who killed his parents and why.
- Opinion
America’s Democrats will now look to 2028
The people of the US have spoken. The loudness and assertiveness with which they have spoken are all out there: the long-beleaguered Democrats have regained the House of Representatives and the Senate, along with a good number of gubernatorial offices, in a splendid landslide.
- Opinion
More From This Author
- Explainer
Draft Cyber Protection Act / Softer penalties, same old concerns
- Long Reads
Hasina’s return: No easy legal route
- Opinion
Women in politics / Needed in protests, not in power
- Opinion
What’s in the name?
- Opinion
Journalist killing / The state cannot escape responsibility for impunity