Delhi's Riot-Control Trucks Gain AI Cameras Without Clear Sanction
Eight armored vehicles bound for Delhi's streets are not merely replacements for aging equipment. According to a report from the Internet Freedom Foundation (IFF), the new Vajra anti-riot trucks being procured by Delhi Police come fitted with pan-tilt-zoom cameras, onboard artificial intelligence processors, and drone-launch capability - turning a blunt crowd-control vehicle into a mobile surveillance platform. The procurement has surfaced just as India's government faces mounting criticism, including from UN experts, over its response to protests tied to allegations of electoral interference.
What the New Vajras Actually Carry
The Vajra has long been a familiar, unsettling presence at Indian protest sites: a steel-meshed truck with a tear-gas launcher mounted on its roof, positioned before police even determine whether an assembly is lawful. The IFF's reporting on the technical specifications for the new fleet describes something considerably more sophisticated. Each vehicle is to be equipped with a 4K camera on an extendable mast "to facilitate the troops with artificial intelligence," five additional 4K cameras, an Intel i7 computer paired with a graphics processor for "AI analytics applications," a terabyte of onboard storage, GPS, Wi-Fi connectivity, and a deck to launch and operate drones from the vehicle itself.
The order replaces four vehicles that were scrapped - yet the government opted to purchase eight, doubling the original fleet. The IFF says it has found no public record explaining that increase, nor any documentation describing what the cameras will record, where footage will be stored or transmitted, or who will be authorized to operate the drones.
A Procurement Process Short on Transparency
This is the central concern raised by digital rights advocates: not necessarily the technology itself, but the absence of a published legal framework governing its use. The IFF has called on Delhi's police commissioner to withhold the contract award until a formal sanction for eight vehicles is placed on record, and until clear, law-based procedures dictate what the cameras, computing systems, and drones may actually do once deployed. Procurement documents, the organization argues, should not leave such fundamental questions unanswered before bidding even concludes.
This pattern is not isolated. A separate Delhi Police tender issued in July seeks 338 fixed and pan-tilt-zoom CCTV cameras equipped with facial recognition technology, to be installed across the Central and New Delhi districts, along with 47 automatic number-plate readers. The contract carries a budget of roughly ₹15.66 crore - about $1.6 million - and specifies a 30-day retention period for biometric data. The IFF notes that this camera network was ordered built around Ramlila Maidan, a ground the government subsequently designated as the permitted protest site for the youth-led Cockroach Janta Party and its supporters. The specifications reportedly call for facial recognition on every camera in the network, along with "people gathering" detection and built-in microphones designed to feed into the city's broader Safe City surveillance system.
Surveillance Infrastructure Meets Civil Unrest
Facial recognition and AI-assisted monitoring are not new to Indian policing, but their expansion alongside riot-control hardware signals a shift in how authorities approach public assembly. A truck that can deploy tear gas, document a crowd in high-resolution detail, and launch a drone overhead represents a convergence of physical force and data collection that raises distinct legal and ethical questions. Who has access to footage gathered at a protest? How long is it retained, and under what oversight? These are not abstract concerns - they shape whether citizens can exercise rights to assembly without fear of being cataloged by facial recognition databases indefinitely.
The IFF's critique of the CCTV tender echoes these worries, pointing to inconsistencies in the bid documents and unclear figures regarding camera counts and specifications. Taken together, the two procurements suggest a broader pattern: surveillance capability is expanding faster than the legal and administrative scaffolding meant to govern it. The government has indicated it expects to award the five-year CCTV contract imminently, leaving little time for the kind of public scrutiny that transparency advocates say is overdue.
