The Digital Knife: Deepfake Defamation and Bangladesh’s Legal Battle for Truth

01. Introduction

Deepfake law in Bangladesh has become a critical topic as artificial intelligence quietly gives birth to a new kind of digital assault that may damage someone’s honour without leaving a trace. For decades, a photograph was considered proof and a video was considered undeniable evidence, but today, both can lie.

A stolen photograph and a few clicks can thoroughly change someone’s entire identity. AI can make hyperrealistic deepfakes by changing or synthesizing text, photos, video, and audio, which commonly appear to be real. In 2017, a moderator on the Reddit forum “Deepfakes” created the first deepfake by putting pornographic films featuring celebrity faces on adult actors’ bodies.^[1]

02. The Deepfake Impact

Social media platforms, such as Facebook, Instagram, and Twitter, have developed into straightforward tools for rapidly spreading material with strong opinions, either favourable or negative. To reinforce gender standards and nationalist scripts that portray someone as immoral or unstable, visual manipulations (such as memes, altered screenshots, and deepfakes) attempt to arouse feelings of shame, disgust, and moral outrage.^[2] Fraudsters currently use artificial intelligence to manipulate stocks and commit other financial crimes, endangering society. They already mimic bank employees over the phone by using artificial intelligence (AI)-generated fake audio to ask for unexpected cash transactions, bank account OTPs, and bank card details.^[3] Soon, artificial superintelligence will be able to mimic live video calls, endangering more lives and causing more damage.

A news media company can acquire a competitive advantage by breaking news first and obtaining video footage from witnesses. In the rush to win a race, people often fail to verify the authenticity of videos of protests, accidents, fake speeches, and other events, including misleading captions that may cause panic elsewhere.^[4]

Syeda Rizwana Hasan, former advisor to Bangladesh’s Ministry of Environment, was targeted in early 2025 when a doctored image – placing her face onto a body sourced from an adult website – was widely circulated on social media platforms through an account called “Chemical Ali”, known for repeatedly targeting prominent women.^[5] As the Ministry of Information and Broadcasting recently reported, it identified 137 fake Facebook accounts, pages, and groups impersonating prominent individuals; 16 misinformation-based websites designed to resemble legitimate news outlets; more than 300 misleading Facebook accounts; and 100 accounts on X involved in spreading false information.^[6]

03. Legal Framework and Deficiencies in Bangladesh

In Bangladesh, individuals can face civil and criminal consequences for acts of defamation. One can submit a defamation complaint to initiate legal proceedings regarding deep-fake offences.^[7] However, in Bangladesh, the concept of defamation is often misapplied. According to section 499 of the Penal Code 1860, any person who, by spoken or written words, or through signs or visual representations, makes or disseminates any statement about someone with the intent to cause harm, or with knowledge or reason to believe that such statements will damage the individual’s reputation, is considered to have defamed that person, except in specific circumstances outlined thereafter.

Anyone found guilty of defaming another individual may face simple imprisonment for a duration that could reach up to two years or with a fine, or with both, under section 500 of the Criminal Code 1860. However, most defamation lawsuits in Bangladesh are primarily initiated for harassment. As a result, defamation cases are often dismissed in lower courts at quite low expense.^[8]

To prevent the misuse of crimes associated with defamation, there was an additional law, the Digital Security Act of 2018. However, this legislation fails to prevent the exploitation of defamation offences. The Cybersecurity Act (CSA) 2026 has been enacted in Bangladesh, repealing the Digital Security Act and the previous Cybersecurity Act. It creates bodies such as the National Cybersecurity Agency to address cybercrime and empowers authorities to rapidly remove harmful AI-generated content and misinformation from social media platforms.^[9] This act lacks explicit procedural safeguards. There is no statutory requirement for the state to alert the content creator, no mechanism for a pre-enforcement adversarial hearing, and no recourse for accelerated judicial review.^[10]

The unfortunate truth is that neither privacy nor data protection is under control. Article 33(b) of our Constitution guarantees the right to privacy. If this right is violated, he or she may file a case under Article 102(1) in the High Court Division. Section 7 of the Right to Information Act of 2009 states that everyone has the right to have their data securely protected.

The Pornography Control Act of 2012 applies when deepfake pornography, nudes, or movies are created utilizing AI technology. However, the effectiveness of these measures may depend not only on the existence of legal provisions but also on how well they are implemented and understood. Potential obstacles include a lack of knowledge about legal remedies, procedural complexity, delays in handling complaints, and uncertainty about enforcement.^[11] The Bangladesh Cybersecurity Strategy (2021-25) is one such initiative. However, there is a disparity between technology adoption and regulation due to a lack of AI-specific legislation and institutional expertise.^[12]

04. Global Response

In 2018, U.S. Senator Sasse introduced the Malicious Deepfake Prohibition Act, followed by Representative Yvette D. Clark’s DEEP FAKES Accountability Act in 2019.^[13] Although these bills failed, President Biden initiated task forces to address cybercrimes in 2022. In the UK, the Online Safety Act was passed in October 2023, criminalizing the non-consensual distribution of intimate deepfakes.^[14]

Conversely, China has a strict ban on certain deepfakes but still faces challenges, such as face-swap fraud and AI-generated content impersonating individuals, as seen in a 2023 case involving influencer “Carola Ilai.”^[15] Zao, a well-known Chinese software, allowed users to upload their own images and swap out their faces for those of their favourite movie characters. After discovering that the rules of service granted users “perpetual and transferable rights to the data posted” and created significant privacy concerns, WeChat banned the face-swapping software Zao.^[16]

The suspected use of pornographic video and audio deepfakes, which entailed posing as a police officer to threaten and extort money from an elderly citizen, was the subject of an ongoing investigation by Indian police in 2023.^[17] To escape social disgrace, this fragile elderly man considered suicide. A European Union policy document outlines various harms from reproduced deepfakes, including sextortion, defamation, identity theft, reputational damage, fraud, election manipulation, erosion of trust, and damage to the justice system and democracy.^[18]

05. The Barriers

Research indicates that most individuals, including experienced social media users, struggle to detect fake videos, particularly high-quality deepfakes. Studies highlight that while some people can spot low-quality fakes to a degree, the overall ability to identify convincing AI-generated videos remains limited.^[19] The challenges countries face include cross-border jurisdiction, identifying malicious actors and authenticating evidence, IP and copyright rights, personality rights related to post-mortem privacy, the need for ongoing research and development, reassessing privacy and data protection, and cybersecurity resources and awareness.^[20]

OpenAI’s support website states that the organisation does not guarantee data confidentiality or security, highlighting vulnerabilities in chatbots that could be exploited by attackers to spread malware or steal confidential information. The unregulated use of chatbots reduces accountability and transparency. While tools like ChatGPT have potential benefits for education, their misuse poses risks to innovation, transparency, accuracy, and reliability.^[21]

06. The Road Ahead

To enhance user trust in AI-driven fake news predictions, developing explainable AI credibility indicators is vital. These indicators can aid users in identifying credible news. However, further efforts are required to tailor these indicators for low-literate, digitally novice social media users, particularly in non-English languages.^[22]

Ensuring compatibility with local language content and providing accessible explanations of AI decision features is crucial. Suggested methods include using colored backgrounds to differentiate fake video types and indicating confidence levels in predictions, aiding comprehension for low-literate users. Efforts to improve media literacy teaching in communities and schools are essential. Governments and communities should educate the public about the harmful effects of technology and social media, emphasizing caution, especially among minors.

Additionally, technology and media companies need to invest in detection tools for deepfakes, enabling viewers to distinguish between real and fake content. Law enforcement agencies should enhance their digital forensic capabilities with machine learning to detect tampering in recordings, verify content authenticity, and trace the origins of AI-generated material.^[23] Additionally, there is a need for establishing victim-oriented legal aid mechanisms.

Effective communication should highlight three core strategies: prevention, detection, and repair, making the subject more tangible. Additionally, Bangladesh should adopt a multifaceted approach that includes fast reporting, public literacy, platform accountability, and enhanced local detection capabilities. The uploader should bear the burden of proof rather than the victim. Bangladesh’s sociocultural reality is marked by instantaneous humiliation before content authentication. Prevention should be both cultural and technical. Before reacting to potentially harmful audio or video snippets, media, community leaders, and families must be trained to demand cryptographic or platform-verified authentication. Additionally, it is recommended to establish decentralized “First Responder” Networks at community and campus levels, which would effectively report and expedite the removal of non-consensual synthetic media from social media platforms before it can spread widely. We must remember that, while AI has militarized the pixels, we still have complete control over the narrative and can restore our digital sovereignty.

⚡ 3rd Cohort: Registration Live

2026 Legal Research Professional Certificate

Lifetime Membership Verified Certification Legal Research Training Publication Opportunities Official References for Jobs

Secure Your Place Today →