Why Delhi Police’s ₹15.7-crore facial recognition CCTV tender is facing scrutiny amid CJP protests: Concerns explained

The tender proposes a city surveillance system rather than just a set of standalone cameras. Delhi Police would install and operate 338 cameras at selected public locations across parts of the capital under a five-year contract. The project matters because several sites are used for public gatherings and protests, raising questions about how monitoring will be controlled.

The equipment includes fixed cameras, pan-tilt-zoom cameras and automatic number plate recognition cameras. The tender also covers video management software, storage hardware and related infrastructure. Proposed sites include Ramlila Maidan, Turkman Gate, Rajghat, Mandi House and the Patiala House court complex.

The system is part of a modernisation plan that police officials say began around two years ago and is being implemented across several states. However, the Internet Freedom Foundation flagged privacy, technical and procurement concerns. The tender does not clearly explain facial-recognition procedures, databases, retention rules or protections against false matches.

Facial recognition technology, or FRT, uses a camera image of a person’s face and compares it with stored facial images. Its purpose is to find possible matches. In this tender, FRT would be connected to a broader camera network, allowing police to monitor selected public spaces and check faces against specified databases.

The article gives a concrete example from the July 2026 NEET-UG paper leak protests at Jantar Mantar. Delhi Police said they used FRT and other technical tools to scan crowds and check individuals against criminal databases. The proposed system would also include fixed, PTZ and ANPR cameras, plus video management software and storage.

The tender does not clearly identify the databases that would be used. It also lacks a standard operating procedure, privacy impact assessment, accuracy thresholds and adequate safeguards against false matches, according to the IFF. That uncertainty is important because a camera-generated match is only an indication, not proof of wrongdoing.

The proposed system is sizeable because it combines several types of surveillance equipment across public locations. The five-year contract covers 338 cameras, alongside the software and hardware needed to collect, manage and store video. Its scale allows monitoring to extend beyond one venue or one type of activity.

The breakdown is 248 fixed cameras, 43 pan-tilt-zoom cameras and 47 automatic number plate recognition cameras. Fixed cameras watch a set view, while PTZ cameras can be directed across a wider area. ANPR cameras focus on vehicle number plates. The project also includes video management software, storage hardware and related infrastructure.

The IFF found inconsistencies in the tender’s stated quantities of PTZ and ANPR cameras. It also flagged technical specifications that appeared copied from older tenders and inaccurate geographical coordinates, including one camera location allegedly placed about 40 kilometres east of Delhi. These issues have added procurement and implementation concerns.

The tender drew attention because its proposed surveillance locations overlap with places used for demonstrations. The IFF noted that the tender was uploaded five days after protests at Jantar Mantar over the NEET paper leak concluded. On 11 October, Delhi Police also de-notified Jantar Mantar as a designated protest site and redirected future demonstrations elsewhere.

Those alternatives included Ramlila Maidan, Burari Ground and CBD Ground in Shahdara. The surveillance tender lists several public gathering sites, including Ramlila Maidan’s VVIP gate, Mandi House and other central Delhi locations. The IFF said the police had chosen protest grounds they had already chosen to watch with facial recognition.

Delhi Police officials said the procurement was part of a modernisation programme initiated around two years ago and was not connected to recent demonstrations. The IFF also said it found no evidence that the tender and the relocation decision were coordinated. The timing and overlap nevertheless intensified scrutiny.

The main concern is that the tender proposes powerful monitoring without clearly describing the safeguards governing it. The IFF said the documents lack a standard operating procedure for facial recognition and do not include a privacy impact assessment. They also fail to explain how sensitive information would be handled and retained.

The tender does not clearly identify the databases against which captured faces would be checked. It also does not specify accuracy thresholds or adequate safeguards against false matches. Without those details, the documents do not show how police would test the system, correct errors or limit the consequences of an incorrect identification.

These gaps matter because the proposed cameras would operate in public places, including protest and gathering sites. The IFF also identified 28 alleged procedural defects, such as inconsistent camera quantities, copied technical specifications, outdated independent-monitor references and inaccurate coordinates. Together, the concerns cover privacy, accuracy and procurement reliability.

Scanning protesters can cause people to be flagged and checked against police databases, even when the technology only produces a possible match. That can bring people with alleged past records to police attention during a demonstration. It can also affect how authorities view a crowd, especially when the system operates in a politically sensitive public gathering.

During the July 2026 NEET-UG paper leak protests at Jantar Mantar, Delhi Police said FRT and other technical tools identified 2,873 people with potential criminal antecedents. Police said 989 allegedly had past records involving murder, robbery or Arms Act violations. These figures describe database-related identification, not proven conduct at the protest.

A person appearing in a criminal database does not establish that they committed an offence at that event. The article specifically notes that the police-reported figures do not, on their own, prove protest crimes. That is why accuracy standards, human checking, clear procedures and safeguards against false matches matter.

Government oversight matters because surveillance in public spaces can record and analyse people while they attend protests, courts, roads or other gatherings. These places are also where people exercise rights such as assembly. Clear oversight helps define what police may collect, why they may use it and how long sensitive information may be retained.

This tender illustrates the need. The proposed sites include Ramlila Maidan, Mandi House, Rajghat and the Patiala House court complex. The IFF said the documents lack a standard operating procedure, privacy impact assessment, clear database rules and safeguards against false matches. It also identified technical and procurement inconsistencies.

Oversight can require transparent procedures, accuracy thresholds, retention limits and checks on access to data. It can also provide accountability when a person is wrongly flagged. Delhi Police say the project belongs to a wider modernisation plan, while scrutiny continues because the network would operate in public spaces, including locations used for demonstrations.