Flock, a company specializing in automatic license plate recognition and AI-driven camera technology, utilizes gig workers from the Philippines for training its surveillance AI. This practice came to light after an accidental leak revealed that these workers, sourced from platforms like Upwork, are tasked with reviewing and classifying footage, which includes identifying people and vehicles in the United States.
With Flock’s presence expanding across thousands of US communities, their technology is employed by local police for various investigations, including carjackings. Furthermore, law enforcement has accessed Flock’s extensive database for purposes such as monitoring and tracking vehicles without warrants, raising significant privacy concerns.
The company’s reliance on overseas workers for AI training illustrates a common trend in the tech industry, where cost-effectiveness drives companies to outsource labor. However, in Flock’s case, the nature of their business—monitoring public spaces—renders the sensitivity of the footage being processed even more critical.
Flock’s cameras not only capture license plates but also gather details about vehicles and their occupants, which can include clothing descriptions. A patent filed by Flock even mentions the potential for detecting characteristics like "race" in footage. The training process for the AI involves annotating footage, which includes categorizing the makes and colors of vehicles, as well as transcribing license plates.
Some of the leaked material detailed instructions for workers on how to label sounds from the footage—distinguishing between audio cues like gunshots and car accidents. This task required listening to audio clips and selecting options from drop-down menus based on the sounds they identified.
After the leak, access to the panel where this information was displayed was revoked, and Flock declined to comment when contacted by reporters regarding the matter. The situation raises pertinent questions about data privacy, surveillance accountability, and the ethical implications of using gig workers for such sensitive tasks.
