Kenya's Data Labelers Fight for Rights and Recognition as AI Training Workforce

Summary

Data labelers in Kenya are organizing to demand better working conditions, fair compensation, and mental health support after years of underpaid labor training AI systems for major tech companies. Michael Geoffrey Asia, a former data labeler who developed PTSD and other psychological issues from annotating explicit content and operating AI sex bots, now leads the Data Labelers Association (DLA) in pushing for systemic change. The workers, who earn just a few dollars a day despite enabling billion-dollar AI valuations, are drawing parallels to colonial exploitation and calling for solidarity across labor and environmental justice movements. The DLA recently held a major organizing event in Nairobi's arboretum to recruit members and amplify worker testimonies about the hidden human costs behind commercial AI systems.

• Labor organizers are framing data labeling as modern-day colonialism, with workers controlled by opaque algorithms for multinational tech companies

Editorial Opinion

This story reveals a critical blind spot in AI industry narratives about progress and innovation. While tech companies celebrate AI breakthroughs, the workers doing the foundational labor to train these systems are systematically exploited, underpaid, and psychologically harmed with little recourse. The organizing efforts in Kenya represent not just a labor justice issue, but a challenge to the entire economic model of AI development—one that extracts value from vulnerable populations in the Global South while concentrating wealth and power in Silicon Valley. Without addressing the human cost embedded in every AI system, the industry's claims to responsible innovation ring hollow.