NYC MTA is rapidly installing AI video analytics in subways

Black Mirror meet NYC subway – AI tracking software is being deploy by the NYC subway for fare evasion enforcement

NBC news was the first to report this and the MTA expects that by the end of the year, the system will expand by “approximately two dozen more stations, with more to follow,” the report says. The report also found that the MTA lost $690 million to fare evasion in 2022.

These look like big numbers of fair evasion, but its in the scale of the NYC MTA budget. The MTA estimates it lost 690 million in uncollected fares in 2022 meanwhile total operating budget of the NYC MTA is 18.6 billion. So lost fare revenue is ~ 3.7% of its total operating budget.

The MTA spends more than this on overtime (5% of its budget) and is currently spending 18% of its operating budget to service its debt. I don’t think installing AI cameras to try to claw back some small % of lost fares is even a strategy, it seems almost certain this is a security initiative that is self-funding partly by cracking down on low income folks.

Further consider that red light cameras in communities often expensive, and after decades of use its not clear if they even make intersections safer. The private company that NYC MTA is contracting with is not know yet, but it seems unlikely that the money spent on this processing capability would be justified by such a small pool of missed fares.

NYC MTA has a huge security budget & AI cameras will inevitably become a standard tool for security

In 2022 it came out that the NYC MTA was paying 1M per month to a private security firm to place security guards at stations and try to reduce the amount of crime. The president of the MTA was quoted as saying about these guards that:

“They’re really there as a deterrent, and to literally be our eyes and ears,”

Transit Authority President Richard Davey said in response to questioning over the security guards

Instead of expensive human security guards at every station, AI powered surveillance could be used for things like advanced threat detection, automated forced deployment and all sorts of security protocols. This doesn’t just mean the NYC subway will become a micro-surveillance state – they will have the ability to be reading people’s faces for things like sentiment, past criminal record, credit score or other sensitive data.

New video AI analytics capabilities will be easy for the MTA to try

Video AI surveillance systems are already in widespread use globally, for a while London was one of the more modern examples but we have learned China has a “one person, one file” system powered by video AI. According to Reuters China’s video AI systems have the ability to learn and optimize for accuracy automatically as it gathers more data. The system is capable of identifying people from partial views and even masked faces when they have acquired enough data on an individual.

The privacy concerns are real

Use of the software in China is widespread beyond police, schools use it to monitor unfamiliar people around their campus for example. However, we do know that this type of system in major Chinese cities is exactly what allowed the government to attempt to keep people in their apartments during COVID while they faced food shortages and death.

Its not just China, in 2021 the NY times reported:

“The technology London will deploy goes beyond many of the facial recognition systems used elsewhere, which match a photo against a database to identify a person. The new tools use software that can immediately identify people on a police watch list as soon as they are filmed on a video camera.”

In that same piece NY times raised doubt that the matching itself is good quality:

“Last year, an independent review of a police trial found many problems, including its accuracy. Of 42 identifications made by the system, only eight were correct.”