How AI surveillance is managing security at India AI Impact Summit

AI-powered surveillance systems are at the core of security at India's flagship tech event, despite significant crowd disruptions on the opening day.

The India AI Impact Summit 2026 commenced in New Delhi on February 16, showcasing the nation’s rapid progress in AI technologies. However, the opening day at Bharat Mandapam faced major logistical challenges, including overcrowding, long queues, and confusion over entry access.

The large-scale global technology event quickly drew criticism from founders, exhibitors, and visitors. Reports indicated significant crowd management issues, including bottlenecks at entry points, unclear instructions, and last-minute closures of exhibition zones that disrupted movement inside the venue. The increased security measures implemented for high-level attendees exacerbated the operational strain, resulting in long queues and instances of attendees being left stranded outside venue halls for hours.

A general advisory issued ahead of the event had already stated that the Bharat Mandampam Expo Arena would remain closed on the opening day. Despite the advisory, videos shared on X showed large crowds gathering near the exhibition sections, contributing to congestion and amplifying confusion at entry points.

“The crowd management relies on an AI-powered surveillance grid,” says Additional CP Devesh Kumar Mahla. He mentioned that the grid is coordinated through at least 32 control rooms monitoring both the venue and its surrounding areas.

More than 20,000 personnel from the Central Armed Police Forces and local police are using these AI tools to track crowd movement in real time, identify potential threats, and respond swiftly to incidents. This deployment comprised several key technologies: an AI-enabled CCTV network, Facial Recognition Systems (FRS), real-time video analytics, and AI smart glasses utilised by on-site officers, that was also deployed during the January 26 Republic Day security operations.

This AI-enabled system works by feeding live CCTV and wearable camera footage into a central command network, where computer-vision models analyse faces, crowd density and movement patterns in real time. Facial recognition tools compare images against police databases, which then send alerts to control rooms and field officers for quick verification and response.