Bangladesh enacted the Digital Security Act in October 2018 as a replacement for the

Information and Communication Technology Act 2006. The ICT Act had itself been weaponised

under Section 57, which created criminal liability for 'publishing, transmitting or causing

to be published' material deemed 'false and obscene' or likely to 'deprave and corrupt' persons.

Section 57 was quietly reproduced and expanded across DSA sections 21, 25, 28, 29, 31, and 32.

Between October 2018 and January 2023, Bangladesh government data show 7,001 DSA cases were

filed against 21,867 individuals. Sixty percent were filed by government agencies, ruling-party

activists, or security forces. The pattern was not incidental: Section 57 of the ICT Act had

been used in roughly the same way, with journalists, opposition politicians, and social media

users the primary targets.

In 2023, the Hasina government rebranded the DSA as the Cyber Security Act without substantive

reform, retaining the same criminal architecture. The interim government's May 2025 ordinance

is the first legislation to actually remove specific provisions. However, the reform is partial.

What the Cyber Security Ordinance 2025 Changed and What It Kept

The CSO removed nine sections that had enabled the most politically targeted prosecutions:

penalties for spreading 'defamatory propaganda' about the liberation war or national figures,

for publishing 'offensive or threatening' information, and for identity impersonation. These

removals were material. They eliminate the specific instruments that generated the bulk of

journalist prosecutions under DSA and CSA.

What the CSO kept is equally important. Section 26 prohibits content constituting communal

hate speech that 'creates anxiety'. The phrase 'creates anxiety' has no defined threshold and

no precedent in Bangladesh jurisprudence. It is structurally identical to the 'likely to

cause deterioration in law and order' standard that courts in comparable jurisdictions have

repeatedly struck down as overbroad. Section 26 also criminalises 'spreading rumors', a term

that remained in the enacted text despite civil society objections during consultation.

The CSO introduced one genuinely novel provision: recognising internet access as a legal right,

and criminalising AI-generated content used for blackmail, sextortion, and impersonation.

Bangladesh became the first South Asian state to enact AI-specific cyber offences. The

implementation gap is significant: no regulatory body, evidentiary standard, or safe-harbour

provision accompanies the AI-crime sections.

The RSF Benchmark: Three Readings of a Volatile Trend

The RSF trajectory captures the political moment more than structural media conditions.

The 2024 rank of 165th reflected the late-Hasina consolidation of media ownership among

politically connected conglomerates and mass DSA prosecutions. The 2025 improvement to

149th measured the initial liberalisation after the August 2024 political transition:

DSA cases were stayed, some journalists released, and the ordinance reform process began.

The 2026 return to 152nd reflects BNP's early governance record: new arrests for social-media

criticism of the government, continued detention of journalists on pre-transition charges,

and no structural change to media ownership concentration.

The RSF's 'political context' subscore drove the 2026 downgrade. This is the RSF dimension

most sensitive to executive behaviour: it measures independence of public media, political

interference in private media, and tolerance for journalistic investigation of state actors.

None of these dimensions improve automatically with a change of government.

The Permanent-Law Vacuum Under BNP

The CSO's legal status as an executive ordinance matters for institutional reasons beyond

legislative procedure. An ordinance issued by an interim advisory council has no parliamentary

record: no committee hearings, no recorded debates, no official interpretation of ambiguous

provisions. When Section 26 is eventually litigated, courts will have no legislative history

to guide construction.

The BNP government's Jatiya Sangsad, seated after the February 12, 2026 election, has not

tabled a permanent cyber security bill. State Minister for Information and Broadcasting

Yasser Khan Chowdhury has made no public statements on a legislative timeline. RSF

recommendations issued in February 2026 called explicitly for a permanent law replacing

the CSO, with press-freedom-compatible safeguards and judicial oversight of surveillance

powers. Those recommendations remain unanswered.

The absence of a parliamentary process also means civil society and press organisations

have no formal channel to contest the retained provisions of the CSO. The Tech Global

Institute and allied civil society groups issued a joint statement in May 2025 documenting

specific remaining concerns; the BNP government inherited both the ordinance and the

outstanding civil society objections.

Journalist Safety: The Operational Dimension

The RSF benchmark and legislative analysis describe the framework. The operational reality is

documented in incident data. RSF counted more than 130 journalists named in 'baseless cases'

since the August 2024 transition. CPJ and allied groups placed the total at 388. The

discrepancy reflects scope: RSF assesses credibility of charges; CPJ counts all named cases.

Both figures exclude journalists who faced extrajudicial harassment without formal charges.

In April 2026, HRW documented four arrests for social-media content deemed critical of the

BNP government, characterising them as a 'continuation of the previous government's

repressive practices'. CPJ separately called for release of four senior journalists

(Farzana Rupa, Shakil Ahmed, Mozammel Haq Babu, and Shyamal Dutta) detained for more

than 18 months on murder charges without charge sheets.

The pattern indicates that the legal instrument driving journalist suppression has shifted

from DSA/CSA criminal provisions to other criminal charges (murder, sedition, contempt)

that operate independently of the cyber-speech framework. This means CSO reform alone

cannot resolve Bangladesh's press-freedom deficit.

Policy Implications

Three legislative actions would materially improve Bangladesh's press-freedom standing before

the next RSF cycle. First, parliament should pass a permanent Cyber Security Act that repeals

the CSO, codifies the nine removed provisions as permanently abolished, and replaces Section 26

with a narrowly defined incitement standard requiring proof of intent and imminent harm.

Second, a broadcast and print media ownership transparency law should require public disclosure

of beneficial owners for any media outlet with more than 5 percent national audience share.

Third, a journalist protection protocol, modelled on Colombia's National Protection Unit

mechanism, should establish a rapid-response legal defence fund and case-tracking register

accessible to courts.

The BNP government's credibility on press freedom will not be determined by the ordinance

it inherited. It will be determined by the permanent law it chooses to write.