4 laws Bangladesh needs.
Three proposed statutes and an AI strategy. These are our recommendations, not enacted law.
Proposed lawOnline Safety Act
One law for online harms and cybercrime, built on platform duties rather than broad criminal bans.
The problem today
Laws on online harm are fragmented and outdated. They lean on broad criminal offenses, vague definitions, executive discretion and reactive takedowns, while doing little about platform design, transparency and accountability.
What we propose
We propose one rights-respecting law covering harmful content and cybercrime together, with duties matched to each platform’s type, functions and risk instead of broad criminal bans.
- User-generated content platforms
- Streaming and subscription services
- Online intermediaries
- App stores
- Messaging services
- AI-enabled systems
- Recommender systems
- Hosting providers
What it would include
Duties of care that scale
Platforms would assess, prevent and respond to serious online harms, in proportion to their size and risk.
Notice-and-action with safeguards
A clear, time-bound process for acting on lawful notices about unlawful content, with counter-notices and protection against abusive takedown requests.
Ways to report and appeal
Accessible, multilingual ways to report harm, get remedies, contest moderation decisions and appeal.
Public transparency reports
Regular public reports on content moderation, recommendation systems, advertising, complaints and government requests.
Stronger protection for those most at risk
Enhanced protection for children, women, journalists, minorities, LGBTQ+ communities and persons with disabilities.
An independent regulator
An independent, technically specialized authority to oversee compliance, conduct audits and impose penalties.
Cybercrime, defined precisely
Clear offenses for hacking, ransomware, phishing and identity theft, with legal protection for good-faith security research.
What would change for you online
- You could report harm, contest a moderation decision and appeal.
- Platforms would publish what they remove and what the government asks of them.
- Organizations would have to report cyber incidents and data breaches within set time limits.
Proposed lawRegulation of Investigatory Powers Act
One law for surveillance and interception, with a judge’s approval first and layered oversight.
The problem today
Surveillance and interception powers are spread across several statutes, licenses, government orders and agency mandates that are not fully public, without clear legal thresholds, independent authorization, oversight, transparency or redress.
What we propose
We propose an independent national review and public consultation, then one law that brings these powers together and balances state security with fundamental rights.
- 1Prior judicial authorization
- 2An independent surveillance commissioner
- 3A parliamentary committee
- 4Periodic independent audits
- 5Annual public transparency reports
What it would include
A judge approves first
Interception and device searches would need prior authorization by an independent court, limited to serious offenses and genuine national-security threats.
No bulk surveillance
Bulk, indiscriminate or untargeted surveillance would be prohibited except under an exceptional framework with heightened safeguards.
Limits on what is kept
Strict time limits, data minimization, retention and deletion rules, and audit trails for everything collected.
Protected communications
Higher thresholds for legally privileged, journalistic, parliamentary and medical communications.
Written orders only
Companies would assist only under lawful, written, reviewable orders; informal or politically directed requests would be prohibited.
Rules for spyware
Buying and using spyware, facial recognition and similar tools would need legal authorization, audits and limits on use against journalists and activists.
What would change if you are monitored
- You would be told about surveillance once it no longer prejudices an investigation, and could complain, seek judicial review and compensation.
- Evidence obtained unlawfully would be excluded.
- Aggregate statistics on surveillance orders would be published.
Proposed lawDigital Commerce Act
One platform-aware law for buying and selling online, from marketplaces to social commerce.
The problem today
Digital commerce relies mainly on policies, guidelines and voluntary codes without binding force. It is unclear whether selling through Facebook, Instagram, WhatsApp, TikTok or Telegram is covered.
What we propose
We propose a technology-neutral, platform-aware digital commerce law, with obligations matched to how each kind of platform works.
- B2CBusiness to consumerAmazon, Daraz
- B2BBusiness to businessAlibaba
- C2CSocial commerceFacebook Marketplace, TikTok Shop
- C2BConsumer to businessUpwork, Shutterstock
- B2B2CSeller ecosystemsShopify-enabled marketplaces
- Super-appsMobile commercebKash, Uber, Pathao
Facebook Marketplace mainly facilitates advertising and offline deals; TikTok Shop runs payments, recommendations, logistics and fulfillment on the platform.
What it would include
Clear rules for online contracts
Rules on electronic contracts, digital signatures, terms of service, returns, refunds and statutory guarantees.
Reach to offshore platforms
Narrow rules for offshore platforms serving users in Bangladesh, while avoiding jurisdictional overreach.
Guards against scams and tricks
Safeguards against fraud, counterfeits, deceptive pricing, dark patterns, fake reviews and influencer marketing abuses.
Know who is selling
Seller verification, traceability of commercial actors and disclosure of sponsored advertising.
Fast, low-cost complaints
Online dispute resolution with response deadlines, transparent refunds and low-cost redress for consumers and small businesses.
Smoother cross-border trade
Digitized customs, simpler procedures for low-value shipments and interoperable payment and logistics systems.
What would change when you buy online
- You would know who you are buying from.
- Complaints would get answers within set timelines, with transparent refunds.
- Fake reviews, dark patterns and AI-generated scams would be targeted.
Proposed strategyAI Strategy
Not a rigid AI law yet: a flexible, principles-based strategy first.
The problem today
AI is entering courts, policing, health care, education, finance and public administration, bringing opportunities alongside serious risks to rights, democratic governance and public trust.
What we propose
We propose avoiding rigid technology-specific legislation for now. Start with a national strategy, ethical guidelines, technical standards, regulatory sandboxes and sector codes, then consider statutory regulation.
- ProhibitedUnlawful biometric mass surveillance, social scoring, discriminatory profiling, unlawful predictive policing
- High-riskImpact assessments before use in sensitive sectors
- Limited-risk
- Minimal-risk
What it would include
Risk-based rules
Rules that distinguish prohibited, high-risk, limited-risk and minimal-risk AI systems.
Bans on the most harmful uses
Prohibit or strictly limit unlawful biometric mass surveillance, social scoring, discriminatory profiling and unlawful predictive policing.
Check before deploying
Human rights, privacy and safety impact assessments for high-risk AI before it is used in sensitive sectors.
A human in the loop
Human oversight, explanations and a way to challenge AI-assisted decisions about jobs, health care, education, credit or housing.
Label AI and deepfakes
Tell people when they interact with AI or AI-generated content, and label deepfakes and synthetic media.
Build local capacity
Support open research, local-language AI, accessibility technologies and digital inclusion.
What would change when AI decides about you
- You would be told when you are dealing with AI, and deepfakes would be labeled.
- You could challenge an AI-assisted decision about your job, health care, education or credit.
How others do it
Comparative Laws, by proposal.
Referenced to show different approaches, not as models for adoption.
Open a proposal to see the laws, grouped by jurisdiction and linked to their texts.
For an Online Safety Act26 laws from 5 jurisdictions
By jurisdiction
Australia 8
European Union 7
- Digital Services Act
- Artificial Intelligence Act
- Directive on Combating Violence Against Women and Domestic Violence
- Audiovisual Media Services Directive
- Regulation on Dissemination of Terrorist Content Online
- Regulation on Political Advertising Transparency and Targeting
- Code of Conduct on Disinformation
Singapore 6
United Kingdom 3
For investigatory powers19 laws from 6 jurisdictions
By jurisdiction
Australia 5
European Union 3
New Zealand 3
For digital commerce18 laws from 5 jurisdictions
By jurisdiction
European Union 7
United Nations 2
United Kingdom 4
United States 4
- Integrity, Notification, and Fairness in Online Retail Marketplaces for Consumers Act
- Guides Concerning the Use of Endorsements and Testimonials in Advertising
- The Electronic Signatures in Global and National Commerce Act
- Strategic Plan 2022 - 2026 on Innovation, Equity, and Resilience: Strengthening American Competitiveness in the 21st Century
For an AI strategy26 laws from 13 jurisdictions
By jurisdiction
Australia 2
Canada 3
Council of Europe 1
European Union 3
Global Partnership on AI 1
New Zealand 1
OECD 2
United Kingdom 2
United Nations 1
United States 6
- National Artificial Intelligence Initiative Act of 2020
- Algorithmic Accountability Bill
- Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile
- Executive Order on Promoting the Use of Trustworthy Artificial Intelligence in the Federal Government
- Executive Order on Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence
- Executive Order on Maintaining American Leadership in Artificial Intelligence