New York City has begun testing an AI-based surveillance system in subway stations with the goal of strengthening security and reducing crime. This initiative is being led by the Metropolitan Transportation Authority (MTA), which confirmed that the technology is already being used at some points in the largest public transportation system in the United States.
The system analyses real-time images to identify behavioural patterns that could pose risks.
Michael Kemper, MTA security chief and former chief of the New York Police Department’s (NYPD) Transit Bureau, is spearheading the project. The AI, Kemper explained, allows law enforcement officers to be quickly alerted when erratic movements, confrontations, or other actions deemed out of the ordinary are detected.
These notifications are intended to facilitate an immediate response from authorities before an incident escalates.
The surveillance system does not use facial recognition or collect personal data. According to authorities, the approach is solely aimed at identifying behaviours, which seeks to protect user privacy.
The initiative comes amid a rise in certain types of crime on the subway, particularly assaults, while major crimes overall declined in 2024. State authorities consider this tool part of a comprehensive security strategy, which also includes a reinforced presence of police and military forces.
How does the AI system work on the New York subway?
The new AI surveillance technology implemented by the MTA analyses images captured by cameras installed in subway stations and cars. The goal is to identify unusual behaviour in real time, without focusing on individuals’ identities.
According to NBC New York, when the system detects an action that matches its alert parameters—for example, sudden movements or physical confrontations—a notification is automatically generated for security teams.
Currently, about 40% of the subway system’s cameras are monitored by staff in real time. AI seeks to expand that capacity without increasing staff, allowing for broader and more continuous coverage, the MTA said. The agency has not revealed the name of the software provider or the specific criteria the AI uses to classify behaviour as risky.
The system is in the pilot phase and is limited to select stations. Its expansion will depend on the results obtained during this trial period. Michael Kemper explained that the tool does not replace human officers, but rather acts as a complement to optimise surveillance and reduce incident response time.
What other security measures has New York taken on the subway?
In addition to the use of artificial intelligence, New York State deployed a contingent of 1,000 National Guard members to assist transit police in late 2024. This action was ordered by Governor Kathy Hochul in response to the perception of insecurity on public transportation, The New York Times reported.
The intervention includes random backpack checks, increased patrols at key stations, and escorting riders during peak traffic hours. These measures are part of a comprehensive security strategy that seeks to reduce incidents and strengthen public confidence in the transportation system, according to state authorities cited by the outlet.
In parallel, citizen awareness campaigns and collaboration with community organisations have been maintained to improve the relationship between police and riders. The MTA noted that security efforts are under constant evaluation and that the use of technologies such as AI will not replace human contact or traditional policing.
What concerns does AI use in public spaces raise?
The implementation of the AI system has drawn criticism from civil rights organisations. The New York Civil Liberties Union (NYCLU) expressed concern that surveillance algorithms could reproduce biases or generate false positives that lead to unjustified interventions, according to CNN.
While the MTA maintains that facial recognition is not used and personal data is not recorded, the NYCLU has called for transparency in the program and the establishment of independent review mechanisms.
These groups believe the use of AI in public surveillance should be governed by clear protocols that ensure the protection of individual rights.
The MTA announced that it will subject the program to technical and legal evaluations during the pilot phase. In addition, external committees are expected to review its performance in relation to legal safeguards. This oversight process was announced in an official notice published on the New York State website (ny.gov) in early May.
What other cities are using AI for public transport safety?
The use of artificial intelligence in transportation systems isn’t exclusive to New York. Other cities like Los Angeles and Chicago have also adopted AI-based tools to improve the safety and efficiency of their networks.
In Los Angeles, for example, weapon-detecting cameras have been tested, while Chicago has used AI to issue tickets for traffic violations, according to Bloomberg.
These experiences reflect a growing trend toward the automation of surveillance in urban environments. However, the effectiveness and social impact of these technologies vary depending on the context and legal framework of each jurisdiction.
In all cases, experts agree on the need to evaluate these programs with verifiable data and under ethical standards.