Recently, Dhaka Metropolitan Police (DMP) issued a public notice regarding traffic case warnings based on surveillance camera footage. For the very first time in Bangladesh, AI-based e-traffic prosecution is being integrated into the institutional mechanisms of traffic law enforcement. Within the first week of introducing this system, a total of 300 traffic cases were filed. Citizens have also welcomed the new AI-powered system.
Traffic rule violations have remained a persistent issue in Dhaka, unaddressed for many years; and one of the major reasons why the capital is consistently listed amongst the world’s most unliveable cities. As such, a new policy to regulate traffic using artificial intelligence is truly commendable.
This also raises an important question: should Bangladesh develop a legal framework that formally recognises and regulates the use of AI by law enforcement agencies?
Currently, Bangladesh does not have any statute or legislative framework that explicitly defines “AI-generated evidence” or how it can be used during a prosecution, or a detailed framework that will ensure citizens’ rights against automated accusations against them. The relevant laws related to the issue are the Evidence Act, 1872 (amended in 2022) and the Cyber Security Act, 2026. The point of commonality between these laws is the absence within them of a definition for “AI” or “e-prosecution.” The most recent statute, the Cyber Security Act, 2026, which was enacted a month ago, surprisingly did not provide a legal definition of “artificial intelligence” or any other term related to AI regulation.
The e-traffic system in discussion operates entirely under administrative authority, without any statutory backup. The public notice issued by the DMP lacked a legislative foundation.
Articles 31 and 35 of the Constitution of Bangladesh guarantee every citizen’s right to the protection of the law and fair trial. However, the entire process of detection and identification of traffic violations is automated, meaning that there might not be any human intervention during the entire process. This might deprive the accused person of any opportunity to defend themselves until the notice has been served. As a result, a question regarding who will evaluate and ensure the system’s errors, biases, false positives, credibility, and accountability remains unanswered.
Additionally, under the Evidence Act, evidence must be authentic, reliable, and admissible. Digital footage is admissible only if its authenticity is established before the court. This raises questions concerning who will verify the authenticity of digital footage and in what manner. Furthermore, can AI be trusted with reliable levels of accuracy yet? And how will the AI-based system be questioned before the court? These concerns need a comprehensive clarification and soon.
Then there is the burden of proof and the question of who will bear it: the state or the accused? There is no proper answer to this question as there is no law to answer it yet.
This is one of the reasons why many countries have ensured legal safeguards while moving faster on AI-assisted enforcement. For instance, the European Union in 2024 enacted the Artificial Intelligence Act, which defines AI systems used in law enforcement as inherently “high-risk.” Therefore, the law suggests mandating transparency, human supervision, and an independent audit of the system to avoid this risk. In 2022, the United Kingdom introduced the Surveillance Camera Code of Practice, which also mentions similar obligations for public automated systems.
The goal for Bangladesh is not to copy these policies or the legislative framework, but to understand the significance of having a designated legislative framework for AI-based prosecution that supplements administrative actions.
To be clear, the problem is not switching to an AI-based enforcement policy, but allowing a legislative gap surrounding the practicality and accuracy of the automated system. Legitimising AI-driven law enforcement policy will not only help gain public trust but also ensure its sustainability in future.
The DMP should nevertheless be congratulated for initiating a timely initiative but should also ensure that the mode of instructions and overall procedure are communicated in a manner easily understandable to citizens. Moreover, there should be transparency regarding how e-traffic prosecution stores the data of citizens. Predictably, other AI-based law enforcement applications like facial recognition, automated surveillance, and predictive policing could soon be integrated into the system. Hence, the question of legalising AI-driven law enforcement policies will not remain hypothetical for too long.
Pinak Sarkar is lecturer in the Department of Law at International Standard University.
Views expressed in this article are the author's own.
Follow The Daily Star Opinion on Facebook for the latest opinions, commentaries, and analyses by experts and professionals. To contribute your article or letter to The Daily Star Opinion, see our guidelines for submission.
