Daniel Mercer
Flock Safety, a company known for its controversial surveillance technology, has consistently claimed that its automated license plate readers (ALPRs) do not have the capability to recognize or track individuals’ faces. a new artificial intelligence tool may change that narrative, as it reportedly has the ability to identify vehicles and their drivers through various other factors.
This innovative tool, named OS Investigate, is currently undergoing testing with a limited number of law enforcement agencies, according to a recent report from Wired. Journalists from Wired, along with an independent security and privacy researcher known as Buchodi, managed to create a mock-up of the experimental tool’s interface using code obtained from the login portal of Flock’s website, although they could not access the tool directly.
The interface of OS Investigate enables law enforcement officers to search for vehicles within Flock’s extensive network of cameras through natural language queries. The system includes sixty-nine prewritten prompts that officers can modify before submitting their searches. One example of a prompt instructs the system to locate “witnesses based on vehicles most seen in [neighborhood] during [last 14 days] during [daily timeframe].” The investigation revealed that while the system requests officers to provide a justification for their searches, it does not enforce any specific requirements regarding the length or detail of these explanations.
Unlike Flock’s original business model, which involved comparing recorded license plates against lists of vehicles previously implicated in crimes, this new system begins with a specific location, timeframe, or movement pattern, and identifies vehicles accordingly. It also excludes large vehicles such as buses, semi-trucks, business vans, and trailers, according to Wired, thereby concentrating the searches on privately owned vehicles.
A representative from Flock described OS Investigate as “a distinct product from Flock’s license-plate reader technology, created to assist investigators in utilizing the information their agencies already possess,” as summarized by Wired. The spokesperson also indicated that the tool’s features and interface could evolve before its public release.
Flock asserts on its website that its ALPRs do not “identify people by their faces” or “determine who is inside a vehicle.” However, by integrating with 911 dispatch records, arrest logs, and other public safety and commercial databases, the new tool reportedly allows officers to make connections between license plates and their drivers, as well as individuals within the drivers’ social circles.
Flock has faced significant criticism following revelations of misuse of its technology by some officers. A recent report by the Washington Post highlighted that at least fifty officers have been charged with or accused of abusing Flock cameras, with many incidents involving stalking of women. In response to public outcry, which has included widespread vandalism of its cameras, the company announced last week that it would implement new safeguards for its cameras—one of which includes a tool that flags suspicious user behavior, which law enforcement clients will be required to adopt by the end of this year.
This controversy emerges amid a surge in the adoption of new AI technologies by police departments across the United States. Companies like Flock Safety and Axon Enterprise, which market themselves as providers of public safety technology, are promoting new AI solutions as a quick remedy for law enforcement agencies striving to do more in the face of staffing shortages. Yet, without comprehensive federal regulations, these agencies largely operate independently in determining how to utilize the technology in a safe and ethical manner.
Brandon Fletcher
