Spies in the Loop: The Workers Building Surveillance Tech

Governments, military forces, and private companies increasingly rely on AI surveillance tools to keep tabs on people and places: scanning crowds for suspects, deploying drones in conflict zones, and flagging suspicious social media posts. These products rely on two fundamental things: billions of data points and low-paid workers to label and organize the data.

This project will seek out the little-known workforce training facial recognition algorithms, labeling drone images, and transcribing audio files for the global surveillance industry.

Amid an “AI arms race” between the U.S. and China and the acceleration of AI warfare in Ukraine, it’s a boom time for data labeling firms. This project will examine the increasingly close links between the data labeling and defense industries and examine the ways in which they are supporting digital authoritarianism all over the world.

Most importantly, it will aim to make connections between two serious harms: working conditions for low-paid, outsourced workers who are often unaware they are training harmful technologies, and the data subjects who are targeted by the tools they have helped to build.