AI cameras are surveilling the roads in Dhaka, Bangladesh


‘What's also needed is a comprehensive reform of the entire transport management system’ Originally published on Global Voices Advox Two security cameras mounted on a pole. Representative image by user Atypeek Dgn via Pexels. Used under a Pexels License .

For years, there was a familiar routine for those who broke a traffic rule in Dhaka, Bangladesh: a brief standoff with a traffic officer, a word or two of pleading, sometimes a quiet handover of cash (a bribe). Other times, drivers simply slipped past unnoticed. Those days may now be coming to an end.

Since May 7, 2026, artificial intelligence (AI)-powered traffic cameras have gone live on Dhaka’s main roads. Installed by the Dhaka Metropolitan Police (DMP) at key signal crossings and intersections, these cameras run on software officially called the “ AI-Based Road Transport Act 2018 Violation Detection System .” When a driver breaks the rules, a notice lands on the vehicle owner’s phone within hours.

This marks the first time Bangladesh has deployed a fully automated digital traffic enforcement system, one that files cases without any human involvement.

How the AI camera system works

When a vehicle runs a red light, crosses the stop line, or blocks the left lane, an AI camera automatically scans the number plate and generates a case on the spot. The software records a few seconds of video footage, cross-references it with the Bangladesh Road Transport Authority (BRTA) database to identify the vehicle owner, and sends a prosecution notice to their address.

In other words, run a red light or drive the wrong way, and the evidence, along with a formal case notice, goes straight to the vehicle owner. Those who receive notices can also view the video of their own violation online.

Former head of the Police Special Branch, Golam Rasul, wrote in a Facebook post : “During the day, you might be able to talk your way out of a ticket with a traffic sergeant, put in a word, make a call, or if you’re bold enough, drop the name of some influential uncle. But this AI camera has no mind, no feelings, no mercy, and it doesn’t scare easily. It only knows two things: the rules, and your number plate.” DMP has  confirmed that cameras are currently active at five major intersections, and installations are underway in additional areas. Over 1,000 cases in 10 days, thousands more expected

According to the Dhaka Metropolitan Police (DMP) , the sophisticated traffic signals and e-prosecution platform have generated around 1,000 traffic cases since their launch on May 7. DMP aims to achieve fully automated case filing within six months across Dhaka, largely removing the need for manual case filing by traffic sergeants or inspectors. Any violation of traffic or motor transport laws, anywhere in the city, will trigger an automatic case and fine. DMP has announced plans to install cameras at a minimum of 500 points across the capital within the next six months.

There’s another catch worth knowing: regardless of who is driving the car when a mistake is made, the case will be filed against the vehicle’s owner. Whether you drive your own car or hand the wheel to another driver, the penalty points will be deducted from the owner’s driving license.  Under BRTA rules , every driver is given 12 points under their license; once those points run out, the license gets canceled. Mixed reactions on Facebook traffic groups

The Facebook groups Traffic Alert (over 447,000 subscribers) and Dhaka Traffic Update (37,300 subscribers) have been buzzing with debate since the cameras went live.

Many are genuinely pleased. Osman Gani , who works as a private driver, said: “People used to be careless about seat belts, but now that violations automatically trigger cases, everyone is forced to comply. When drivers follow the rules, the road feels a little safer.”

But criticisms are also plentiful. In the Dhaka Traffic Alert group, one user named Ismail Hossain wrote :

AI cameras are catching ordinary people’s cars at signals… fine. But what about local buses picking up passengers in the middle of the road? Rickshaws going the wrong way? Buses belching black smoke without fitness certificates? Transport blocking the road with no designated stops? When does the camera become ‘automatic’ for them? If there’s a law, it should apply to everyone. Catching only private cars and bikes doesn’t make a ‘smart traffic’ system. If the biggest troublemakers on the road are ignored, people won’t question just the AI — they’ll question the entire system.

Another user, Omar Faruk, raised a different concern : “Pedestrians don't follow signals either. While vehicles get a green light, pedestrians still cross, which forces cars near the front to inch forward and stop. If the camera targets those cars, why should the owner bear the blame?”

On a more encouraging note, a user named BrilliantAlligator9993 shared a photo taken at 5:15 a.m. at Banglamotor and Karwan Bazar, showing drivers actually obeying traffic lights at that early hour. The caption read: “Glad to see people have actually listened and are following the traffic lights. Really want to see how DMP Traffic deals with the buses, CNGs [natural gas tankers], and auto-rickshaws.”

The emergence of scammers has also alarmed people. Fraudsters have been calling vehicle owners , falsely claiming they were caught violating traffic laws, and demanding money. DMP has warned the public that no fine should ever be paid to an individual, only through official banking channels.

Another incident briefly went viral: a photo circulating online appeared to show an AI camera stolen from a 300-foot stretch of road. The X account Voice of Gen-Z posted:

দেশের বিভিন্ন স্থানে লাগানো অত্যাধুনিক এআই ক‍্যামেরাগুলো চুরি করে নিয়ে যাচ্ছে একটি মহল।

এটি ঢাকার ৩০০ ফিট সড়কের একটি ক‍্যামেরার দৃশ‍্য। পুরো সিস্টেম খুলে নিয়ে গেছে। pic.twitter.com/KMFsyF9OOL — Voice of Gen-Z (@VoGen_Z) May 11, 2026 Some people are removing and taking away the state-of-the-art AI cameras installed across the country. This is a scene from a camera on a 300-foot road in Dhaka. The entire system has been taken away.

It later turned out the camera had not been stolen at all, it had fallen and broken in strong winds. The object in question was also a traffic light, not an AI camera.

Battery-powered rickshaws: The question the AI can’t answer Rickshaws are a major cause of traffic jam in Dhaka metro. Image via Wikimedia Commons by Towhidul. CC BY-SA 4.0 .

The biggest challenge for Dhaka's AI cameras may not be drivers who disobey the rules; it may be the vehicles the system simply cannot see. org/2024/12/27/how-to-stop-battery-operated-rickshaws-from-causing-new-problems-on-dhakas-roads/">How to stop battery-operated rickshaws from causing new problems on Dhaka's roads Unregistered manual and battery-powered rickshaws flood the city’s lanes and main roads in enormous numbers, operating entirely outside the law with no registration and no license plates. The AI camera system’s entire mechanism depends on scanning plates and matching them to the BRTA database. A vehicle with no plate is, for all practical purposes, invisible to it.

DMP and the government have yet to offer any clear explanation of how these vehicles will be handled under the new system. The question keeps emerging on Facebook traffic groups, and so far, it remains unanswered.

Professor B M Mainul Hossain, Director of the Institute of Information Technology (IIT) at the University of Dhaka, told Global Voices via telephone:

The success of this initiative will depend on a few key things: system transparency, the right to appeal, and above all, a clear policy on unregistered vehicles. AI cameras are undoubtedly an option to explore towards restoring order on Dhaka’s roads. But technology alone is not enough. What's also needed is a comprehensive reform of the entire transport management system.

Dhaka’s roads are now under the watch of AI. The real question is whether that watch will be fair, and whether it will look equally at everyone. Written by Nurunnaby Chowdhury View original post (English)

Published: Modified: Back to Voices