Bangladesh’s new proposed Cyber Law could make dissent a crime

Bangladesh’s proposed new cyber law promises to make the digital world safer. But buried in the draft are provisions that would give the state considerably more power over what Bangladeshis may say online and who may complain about. It would also give power on how quickly the government can make the things that they don’t like disappear.

The danger of these comes from the way several provisions fit together.

Under the draft, an insulting Facebook post can become a criminal offence. An unverified but ultimately true report can potentially be treated as a “rumour.” A meme can attract the same liability whether someone created it or merely shared it. Government agencies can order content removed before a judge sees the case. 

Police can arrest without a warrant for offences that now encompass defamation. Some speech offences can be tried by mobile courts. And a loosely defined category of people with a “legitimate interest” can initiate complaints. Taken together, they create something closer to an enforcement system for regulating political speech.

The most consequential change begins with Section 25.

The existing law, according to the document, confines Section 25 principally to sexual harassment, blackmail, revenge pornography, sextortion and child sexual abuse material. The proposed amendment inserts “defamation,” humiliation and bullying into the same provision. 

Its explanation of humiliation is broad: words, conduct, publications, gestures or actions that diminish another person's dignity, honour, reputation or social standing so that the person appears humiliated, contemptible or dishonoured to others can fall within it. 

The maximum punishment under the proposed Section 25(1), according to the document, rises to five years in prison or 20 lakh taka, from two years or 10 lakh taka under the current law. 

The distinction matters because political speech is frequently designed to diminish the reputation of its target. Opposition politics does this and editorial cartoons do it. Polemics do it. A protest slogan may do it as well as a citizen calling a minister incompetent or corrupt may certainly do it.

The draft does not simply police false factual assertions. Its definition extends to words, conduct and even “gestures.” That makes truth an incomplete safeguard against prosecution. The offence can turn on humiliation itself rather than solely on whether a factual allegation was false.

The draft goes further when the target is a woman or a child. Section 25(3), as described in the document, would allow punishment of up to 10 years in prison or a 40 lakh taka fine for an offence under Section 25(1) committed against a woman or someone under 18.

Protection or what?

The ostensible rationale…stronger protection for women and children online is easy to understand. The drafting is much harder to defend. Because defamation has been inserted into Section 25(1), the aggravated punishment does not appear limited to sexual abuse or exploitation.

Defaming a female politician could therefore expose the speaker to twice the maximum imprisonment applicable when an equivalent statement is directed at a male politician.

The result is an asymmetry: identical political speech can potentially carry a five-year maximum in one case and 10 years in another because of the sex of the person being criticised. The document itself warns that a provision intended to protect women and children from online abuse could thereby become an instrument for political defamation cases. 

There is another expansion hidden in a definition. Under Section 2(1)(p), “person” includes not merely human beings but institutions, companies, partnerships, firms and other legal entities. Section 25 refers to conduct against another “person.”

That potentially gives companies something resembling criminal protection against reputational harm. A furious customer review of a hospital, bank or restaurant could therefore move from consumer speech into the territory of criminal law. The same material could also be removed administratively under the expanded Section 8. 

The treatment of defamation is particularly revealing because the draft borrows from Section 499 of the Penal Code without, according to the document, clearly carrying across its protections.

The draft's Explanation 2 invokes Penal Code defamation and extends it to information that is false, distorted, misleading, insulting or harmful. But the document notes that the recognised exceptions associated with Section 499, including protections involving true statements made for the public good and good-faith criticism of public conduct, are not reproduced. It also adds “insulting” language that is not part of the formulation being imported. 

That omission could matter a lot to journalism.

An investigative story about corruption is almost necessarily harmful to somebody's reputation. A report exposing abuse at an institution may be humiliating. A truthful investigation can be deeply insulting to the person it exposes. Defamation law normally has to distinguish reputational injury from unlawful reputational injury. If the offence is imported while its defences are weakened or omitted, that distinction begins to collapse.

The draft then confronts a form of speech that has become central to Bangladeshi political culture…the meme.

Criminalizing satire and investigative journalism?

Section 25 expressly encompasses graphics and material created or edited using artificial intelligence. Yet the draft, according to the document, provides no explicit exemption here for satire, parody or cartoons. The document notes that exemptions exist elsewhere for journalism, research, education and whistleblowing, demonstrating that the drafters know how to write exceptions when they choose to. 

The omission is important because satire works precisely by distortion. A caricature exaggerates while a meme removes context. A parody puts words into somebody's mouth and an AI-generated image may make a politician appear ridiculous.

Section 27 magnifies the problem by treating assistance in an offence as punishable in the same way as the underlying offence. The document argues that because Section 25 uses the concept of publication or dissemination, somebody who merely shares a meme can potentially face the same liability as its creator. 

Even an approving reaction could become legally problematic because the definition of humiliation encompasses an “indication” or gesture. The practical question for millions of social-media users would no longer be simply, “Can I write this?” It would become, “Can I share it, comment on it or react to it?”

But the draft's most consequential provision for journalism may be the entirely new Section 26A.

It criminalises publishing or disseminating “rumour” and false information in cyberspace. The draft's definition, as reproduced in the document, encompasses unsupported or unverified information, news or claims that create — or merely have the potential to create — confusion, fear, excitement or social instability. 

Three choices in that definition radically expand its reach. Information apparently need not ultimately be false. Being “unverified” can be enough. Actual harm need not occur; the possibility of producing certain effects can suffice. And at least one branch of the provision, as analysed in the document, does not require an intention to deceive.

That creates a strange legal test. Suppose a journalist reports, citing hospital sources, that protesters have been killed before the government confirms the deaths. Suppose the report is proved entirely correct six hours later. Its truth may not solve the initial problem.

At publication, the information was officially unverified. And news of deaths during a protest obviously has the potential to create “excitement,” fear or unrest. The maximum punishment identified in the document is 10 years' imprisonment or a 40 lakh taka fine. 

That reverses the ordinary logic of journalism. Reporting exists partly to establish facts before governments choose to acknowledge them. Investigative journalism routinely publishes information authorities dispute. Reporters rely on documents, witnesses and confidential sources precisely because official confirmation is unavailable.

A law that effectively makes prior verification by authority safer than independent verification by a newsroom does not merely punish false reporting. It changes the incentive structure of reporting itself.

The rational editor confronted with a credible but politically explosive scoop will have to ask a new question: Is this story worth a possible 10-year criminal case?

That is how chilling effects work. A government does not have to convict every journalist. It merely has to make publication sufficiently risky.

Problems for citizen journalism 

Citizen journalism is even more exposed. So are activists documenting clashes, disappearances, police conduct or election irregularities in real time. Their reporting often begins precisely in the period when facts are incomplete and official confirmation unavailable.

Forwarding material can also attract liability. The document reads Section 26A together with Section 27 to conclude that forwarding an unverified whatsapp message may constitute dissemination or assistance, potentially attracting the same maximum punishment. 

The law may even create a paradox for fact-checkers. To debunk a false claim, a fact-checker normally has to reproduce it: a screenshot, quotation or video is shown and then disproved. Yet the draft contains no clear exception identified in the document for republication undertaken for fact-checking or criticism. The very act required to expose misinformation could therefore technically constitute its publication or dissemination.

This is where imprecision becomes a form of power. A narrowly drafted law tells citizens what they cannot do. A broadly drafted law makes them ask officials what they are allowed to do. Section 8 moves that power from prosecution to censorship.

Under the proposed Section 8(2)(c), law-enforcement agencies, the Information and Broadcasting Ministry and other agencies or forces authorised by the government could seek the removal or blocking of information considered defamatory to a person or institution or “disrespectful to the state.” 

“Disrespectful to the state” is particularly troublesome because, according to the document, the term is undefined.

A state is not a person. It does not possess feelings or a private reputation. Democratic government depends on citizens being able to ridicule state institutions, condemn the police, accuse ministries of failure and describe government conduct in language that officials may find profoundly disrespectful.

The draft does not merely give this authority to police. The document says the power is expanded to the Information and Broadcasting Ministry and an open-ended category of “other agencies or forces” authorised by the government.  Judicial supervision comes afterwards.

Under Section 8(4), material may first be removed, transferred or blocked, with approval from the relevant tribunal obtained within three days. In other words, the executive can act first and justify itself later. 

Three days of course is not a minor interval on the internet. It can be the entire useful life of a piece of political information. A video showing violence during a demonstration matters while the demonstration is happening. An allegation about election manipulation matters before voting ends. Restoring lawful material three days later does not restore the speech's political value.

Some transparency 

There is one meaningful safeguard: Section 8(3), according to the document, requires information about blocked content to be made public. That transparency provision should be preserved. 

But transparency after censorship is not a substitute for independent authorisation before censorship. The enforcement machinery is equally consequential.

The current law restricts who can initiate cases, an important response to one of the most notorious features of Bangladesh's earlier digital laws: politically motivated complaints by people with little direct connection to the allegedly injured person.

The draft expands the definition of an aggrieved party to include legal representatives, guardians, family members and any other person with a “legitimate interest.” The document argues that the final category has no clear boundary. In cases involving Section 26A, where the supposed injured party can include the public, an institution or the state, the universe of potential complainants becomes wider still. 

That raises the possibility of something Bangladesh has seen before…one Facebook post producing cases in multiple districts, filed by political supporters claiming injury on behalf of a leader, institution or public interest.

The draft also makes Section 25 cognisable. Because defamation is being inserted into Section 25, police could arrest without a warrant for conduct that includes defamatory online speech. Section 25 remains bailable, an important distinction, but bail does not eliminate the coercive effect of arrest. 

Mobile courts and provision for expatriates

The draft adds another unusual mechanism: mobile courts.

Section 41(3) would permit offences outside a specified category to be placed under the Mobile Court Act. The document says this could include Section 25(1)-(2) defamation and Section 26 offences, moving them away from exclusive adjudication by the Cyber Tribunal and potentially before executive magistrates. Appeals from mobile-court decisions, it notes, proceed through another executive-magistracy channel. 

For ordinary regulatory offences, mobile courts serve one purpose. For political speech, they present a fundamentally different problem. Questions of defamation, satire, intent, digital authenticity, public interest and constitutional expression are not simple matters of catching somebody selling adulterated food at a market stall.

They require evidence, argument and independent judicial scrutiny.

That concern becomes sharper in the age of deepfakes. The draft itself recognises that AI can manufacture and manipulate audio, images and video. Yet the document questions how such evidence will be assessed in an expedited mobile-court environment when the law elsewhere establishes digital forensic mechanisms.

The draft also reaches Bangladeshis abroad. Sections 4(1) and 4(2), according to the document, apply the law to Bangladeshi citizens committing covered offences outside the country and in circumstances involving computers or networks located in Bangladesh. Expatriate journalists, Youtubers and activists therefore do not necessarily escape its reach by leaving the country. 

At the same time that enforcement power expands, accountability for its misuse appears to narrow.

A new Section 52 would protect officials from civil or criminal proceedings for acts performed in “good faith” under the law. Someone whose content is wrongly blocked may apply for relief, but the document notes an asymmetry: victims of cybercrime may receive compensation under another provision, while people harmed by wrongful blocking or accusation receive no equivalent compensation mechanism and officials receive statutory protection for good-faith action. 

Then there is institutional oversight. The National Cyber Security Council is described in the document as the agency's principal supervisory body. The current requirement for it to meet at least once every three months would be changed to once every six months, cutting formal meetings in half precisely as the draft broadens blocking and enforcement powers. 

The most revealing measure of the draft, however, may be its punishments.

Crimes and punishments 

The document places four offences beside one another. Incitement to religious or ethnic violence under Section 26 carries a maximum of two years. Destroying data by hacking critical information infrastructure can bring seven years. Cyberterrorism resulting in death or serious injury can bring 10 years. Disseminating unverified information under Section 26A can also bring 10 years. 

That is an extraordinary hierarchy of harm.

A person who circulates information that has not yet been verified can face five times the maximum prison term identified for directly inciting religious or ethnic violence. The former need not necessarily produce physical injury. The latter, by definition, concerns conduct capable of setting communities against one another.

The central problem with this draft is that it repeatedly collapses distinctions a democratic speech regime depends upon: between falsehood and unverified information; defamation and insult; creation and sharing; abuse and satire; harmful misinformation and reporting in progress; criticism of government and “disrespect” toward the state; judicial restraint and administrative removal.

And it attaches severe criminal consequences to those blurred boundaries.

The likely effect would extend well beyond defendants who eventually appear before a court. Editors would delay stories and reporters would wait for official confirmation. Cartoonists would reconsider caricatures. Facebook users would think twice before sharing political memes. 

Activists would hesitate before uploading footage from a protest and fact-checkers could become cautious about reproducing the very misinformation they are trying to expose.

—

Asif Shahriyar Sushmit is the Executive Director, Institute of Policy Dynamics