Anyone with a little money can buy access to Moscow’s surveillance system of tens of thousands of cameras along and check footage stored over the previous five days.
Sellers on forums and messenger groups that trade illegal data also provide facial recognition lookup services.
To ensure safety in the city, there are over 175,000 CCTV cameras in Moscow, most of them installed at entrances and more than 4,000 present in crowded places.
Back in 2017, the mayor’s office in Moscow stated that facial recognition technology integrated with Russia’s police databases had been activated and it was pulling data from 3,000 cameras.
The plan was to link the rest of the video surveillance to the facial recognition system.
The city’s website says that the video surveillance system can be accessed by “employees of federal government bodies, the Moscow Mayor and their authorized officials, law enforcement and executive authorities.”
Investigative media outlet MBKh Media found that access to this technology and the live streams is being sold on underground forums and chat rooms.
Andrey Kaganskikh, the journalist that did the investigation says that the sellers are law enforcement individuals as well as government bureaucrats that can log into the Integrated Center for Data Processing and Storage (YTKD), the very system that keeps the data from cameras in Moscow.
Whoever wants to check the live stream from a camera receives a unique link to the City CCTV System that connects to all public cameras in Moscow. The URL works for five days, Kaganskikh says.
This is the same period mentioned on the city’s CCTV section for storing footage from crowded places, shops, and courtyards. Data from educational organizations is saved for 30 days.
Furthermore, government officials or police officers sell their login credentials to the system to provide unlimited access to all cameras. The price of admission is 30,000 rubles ($470), according to Kaganskikh.
To test the facial recognition attributes, the investigator provided a photo of him to a seller. The search returned 238 images of people (male and female) with a similar look from 140 cameras, along with a list of addresses and times they were caught on camera.
Among the metadata available for each photo there was a label indicating if the person was regularly seen by that camera or if they were spotted for the first time.
As for the accuracy of the results, none of the photos returned were of the investigator. However, the facial features were similar to the input and the system assessed a similarity of 67%.
Kaganskikh says in a video report that the poor results may be explained by the limited number of cameras connected to the face recognition algorithm.
The investigator says that the sellers he’d been in contact with told him that the police have free access to this technology. No justification, such as a warrant for a legal investigation, is required to run a search.
Finding the sellers is not difficult, either. Apparently, the forums are indexed by search engines, so their content is easy to discover. This type of access is popular with private detectives, scammers, or individuals that trade surveillance services.