Flock, a company specializing in automatic license plate reading and AI-driven surveillance technology, has been revealed to utilize gig workers from the Philippines for the training of its machine learning algorithms. This information surfaced due to an accidental leak, which showed that these workers were tasked with reviewing and classifying footage primarily from U.S. locations.
The revelation raises significant concerns regarding who accesses the footage gathered by Flock’s cameras and the geographical location of the personnel involved in the review process. Flock operates extensively across the United States, deploying its technology in numerous communities used routinely by law enforcement to investigate incidents such as carjackings. Furthermore, local police use this system to assist ICE in various searches.
Employing overseas workers is a common practice in the AI sector, often driven by lower labor costs. However, in the context of Flock’s surveillance business, the sensitivity of the footage collected places this practice under scrutiny.
Flock’s surveillance setup continuously captures data on license plates, vehicle colors, makes, and models, allowing law enforcement to conduct nationwide searches on vehicle movements without requiring a warrant. This has led to lawsuits and concerns about privacy, as demonstrated by a recent case where civil rights organizations sued a city heavily equipped with Flock cameras.
The training output from the overseas gig workers comprises categorizing images of vehicles and individuals extracted from the footage, which includes the potential identification of race based on a Flock patent. The exposed leak also included insights into worker performance metrics related to annotation tasks, where these individuals categorized various aspects of the captured footage.
Additionally, instructions provided to workers outlined how to evaluate audio recordings from the surveillance footage, guiding them to identify specific sounds associated with incidents such as car accidents or gunfire. The nature of these instructions indicates that the tasks are not limited to merely visual identification but encompass auditory analysis as well.
After 404 Media reached out for comment regarding the exposed training panel, the content became inaccessible, and Flock representative declined to provide further statements.
The implications of utilizing international gig workers to analyze domestic surveillance footage raise important questions about privacy, security, and the ethical considerations associated with modern surveillance technologies.
