AI’s Hidden Workers

Recently, U.S. regulators have begun to take notice as a wave of companies have misrepresented their services as AI-powered or fully automated, when in reality those services were dependent on human labor. Investors and regulators have successfully filed claims of securities fraud based on an AI washing theory upon learning that promised AI technology was either rudimentary or in some in-stances, non-existent. Much of the back-office work that powers artificial intelligence (AI) takes place on digital labor platforms in developing countries, including Kenya, India, and the Philippines. Data annotators and content moderators make AI possible, but platform workers often receive meager wages, lack legal protections, and are denied the basic rights and benefits traditionally associated with employment status. Some workers face psychological harm from seeing the worst parts of humanity as they set guiderails for AI and make it safe for the general public to use. While the hidden nature of this work means that it has received scant attention to date, that dynamic needs to change. Although not as straightforward as the investor protection system, there are ways to ameliorate working conditions for AI’s hidden workers. Individual jurisdictions have begun to introduce new legislation aimed at protecting crowdworkers, and the United Nations – International Labor Office is poised to pass a new international standard for digital labor platforms. This article argues these regulatory changes are necessary to protect the rights of AI’s hidden workers.