A new investigation has pulled back the curtain on how a leading U.S. surveillance company builds its AI systems. Flock, the startup behind thousands of automated license plate readers and facial recognition cameras across the country, has been outsourcing the labor-intensive task of training its algorithms to gig workers in the Philippines, according to a report from 404 Media.
The findings, based on a cache of exposed data, show that Flock relies on a remote workforce to manually annotate footage captured by its cameras. Workers were tasked with identifying vehicle colors, makes, and models, transcribing license plates, and labeling audio clips from car accidents. This behind-the-scenes process, often referred to as “AI training,” is essential for the accuracy of Flock’s surveillance tools.
[IMG_1]
Flock’s cameras are installed in thousands of U.S. communities, often maintained by local businesses and municipal agencies. They form a centralized surveillance network that constantly scans license plates and pedestrians, categorizing individuals based on clothing and potentially other characteristics like gender and race. In some cases, local police have used Flock to assist Immigration and Customs Enforcement (ICE) in monitoring minority communities.
The exposed documents did not specify the origin of all annotated footage, but screenshots showed license plates from states including New York, Florida, New Jersey, Michigan, and California. This suggests the data being processed by overseas workers comes from a wide geographic range.
Flock is not alone in this practice. Other tech companies have faced scrutiny for using low-paid international labor to power their products. Amazon’s “Just Walk Out” stores, for example, were found to rely on gig workers in India watching shoppers remotely. Similarly, the AI startup Engineer.ai, which claimed to simplify app development, was exposed for passing off human-written code as AI-generated.
However, the key difference with Flock is the lack of consent. While services like Amazon’s stores are voluntary for consumers, Flock’s surveillance operates without the explicit opt-in of the individuals being monitored. This raises significant ethical questions about who decides who gets watched and who does the watching, with exploitation present on both ends of the system.
As surveillance technology becomes more pervasive, the reliance on low-cost overseas labor highlights a troubling trend in the AI industry. The practice not only raises labor concerns but also questions about the transparency and accountability of these systems.
