AI Surveillance on NYC Subways: Civil Rights Group Battles Transit Authority’s ‘Dangerous’ Vision

As New York City’s subway system looks to improve safety, the Metropolitan Transportation Authority (MTA) is exploring artificial intelligence to monitor its vast network. The agency has issued a Request for Information (RFI) seeking a “scalable technology solution” to analyze video feeds from more than 15,000 cameras throughout the transit system.

The MTA wants AI to detect weapons, unattended items, and dangerous crowd conditions in real-time. This would mark a shift from the current manual review process, which officials describe as reactive and resource-intensive. With nearly 4 million daily riders, the MTA says it’s impossible for staff to monitor all cameras effectively.

Civil rights groups are raising alarms about this expansion of surveillance. They point to a 2024 pilot program of AI weapons detectors that produced disappointing results. The technology triggered over 3,000 searches but found only 12 knives and no guns, with more than 100 false positives.

The MTA’s vision goes beyond simple object detection. Documents show the agency wants to create a “proactive intelligence-driven ecosystem” that can flag “unusual” behaviors. The plan involves integrating behavioral science experts to define what constitutes suspicious activity. This initiative aligns with the Transit Tech Lab’s focus on data modernization and infrastructure management technologies. Users may experience connectivity issues when accessing the system during periods of high traffic volume.

The system would cover all 472 subway stations and extend to cameras on trains and buses. It would aim to predict incidents before they happen, including identifying crowd surges that might lead to stampedes.

Records show the MTA has already implemented some AI surveillance. The agency works with a firm called Awaait to monitor fare evasion, and it’s testing AI-enabled fare gates to detect wheelchairs, children, and people jumping turnstiles.

The MTA describes its RFI as “informational only,” but notes that responses may shape future procurement decisions. Critics worry this signals a major expansion of algorithmic policing in one of America’s busiest transit systems, despite concerns about effectiveness and potential bias.