Md. Parvez, Legal analyst, Bangladesh.
Artificial Intelligence is no longer science fiction. It is already here in Bangladesh, deciding who gets a loan, helping doctors diagnose diseases, and changing how our garment factories operate. As the nation pushes towards its “Smart Bangladesh 2041” goal, AI is seen as a powerful engine for progress. But experts are now warning that without proper laws, this engine could drive us down a dangerous road.
The promise of AI is immense, but so are the risks. The same technology that can create opportunities could also deepen inequality, cause job losses, and make life-changing decisions with no accountability. This has led to an urgent call from tech experts, lawyers, and business leaders for the government to create a legal framework to guide AI’s growth.
AI in Everyday Life: Promise and Peril
For the average citizen, the impact of unregulated AI could be significant.
In the financial sector, AI helps banks and mobile financial services approve loans for people in rural areas who might have been overlooked before. But if the AI is trained on biased data, it could unfairly deny loans to women or farmers, with the applicant receiving only an automated rejection and no one to appeal to.
In healthcare, AI can analyze medical scans with incredible speed and accuracy, helping doctors in remote upazilas make better diagnoses. But what happens if an AI system makes a mistake? Our current laws are designed for human error, leaving a critical legal gap. Patients could be left with no way to seek justice if they are harmed by a faulty algorithm.
Perhaps the biggest concern is for our economy’s backbone: the Ready-Made Garment (RMG) sector. AI-powered robots and quality control systems can make factories more efficient, but they also threaten the jobs of millions of workers, most of whom are women. Without a national plan, we risk mass unemployment and social disruption.
Our Laws Haven’t Kept Up
While Bangladesh has laws like the Digital Security Act and a proposed Data Protection Act, these are not enough. Experts say these laws manage our data—the ‘fuel’ for AI—but they do not regulate the AI ‘engine’ itself.
The law is currently silent on the most important questions: Are AI decisions fair? Can we demand an explanation for them? Who is legally responsible when an AI system causes harm?
A Call for a Risk-Based AI Act
The solution is not to ban AI, but to manage it smartly. The growing consensus is that Bangladesh needs a dedicated AI Act based on risk. This means the level of rules would match the level of danger.
For “high-risk” AI used in healthcare or finance, strict rules would be required, such as mandatory testing for bias and ensuring a human can always override the machine’s decision. For “low-risk” AI, like customer service chatbots, the rules would be simple, only requiring that the user knows they are not talking to a person. Some applications, like government social scoring systems, could be banned entirely.
To move forward, the government is being urged to form a committee with experts from technology, law, and civil society to draft this new law. They also recommend creating an independent AI Authority to oversee its use and investing in programs to retrain workers whose jobs are at risk.
Building a “Smart Bangladesh” is about more than just technology; it is about building a safer and more just society for everyone. By creating smart rules for AI now, we can ensure this powerful technology helps us all, not just a select few.
A Proposed Architecture for AI Law Enforcement in Bangladesh
A successful AI law requires a robust and well-defined enforcement architecture. The following ten components are proposed as essential elements of an effective regulatory regime.
Define Clear Legal Terms and Standards: The law must provide precise definitions for “Artificial Intelligence system,” “high-risk AI,” “transparency,” and “accountability” to ensure legal clarity and prevent ambiguity.
Establish a Regulatory Authority: An independent, specialized commission should be established to oversee the AI sector. Its mandate would include certifying high-risk AI systems, conducting audits, monitoring systemic impacts, and serving as a primary body for receiving public complaints.
Mandate Ethical Impact Assessments & Audits: High-risk AI systems, particularly those used in the public sector or affecting fundamental rights, must be required to undergo a mandatory AI Ethical Impact Assessment prior to deployment. Regular third-party audits should be required to ensure ongoing compliance.
Strengthen Data Protection Laws: As data is the lifeblood of AI, a strong and enforced data protection law is a prerequisite. Individuals must have clear rights to access, correct, and delete their personal data, with significant penalties for misuse.
Impose Transparency and Explainability Requirements: For AI decisions that significantly impact individuals (e.g., in employment, credit, or welfare), the law must mandate “explainability.” Individuals have a right to be informed when they are subject to an automated decision and to receive a meaningful explanation of the logic involved.
Guarantee Human Oversight: The law must stipulate that for critical decisions, particularly in high-stakes domains, a human must retain ultimate responsibility and possess the authority to intervene and override the AI system’s output.
Provide for Remedies, Redress, and Penalties: A clear legal pathway must be established for individuals and groups to appeal automated decisions and seek compensation for harms caused by AI systems. The regulatory authority must be empowered to levy significant penalties on organizations that violate the law.
Invest in Capacity Building and Awareness: Effective enforcement requires knowledge. The government must invest in training programs for policymakers, judges, lawyers, and technologists, alongside public awareness campaigns to educate citizens about their rights in the age of AI.
Ensure Regular Review and Adaptability: Given the rapid pace of technological change, the AI law must include a statutory requirement for regular review and amendment. The use of “regulatory sandboxes” could also allow for the testing of novel AI systems in a controlled environment.
Foster International Cooperation and Harmonization: The regulatory authority should be mandated to engage with international counterparts to align with evolving global norms, harmonize standards where appropriate, and collaborate on addressing cross-border AI challenges.
Discussion about this post