This paper investigates the often-overlooked human labour underpinning the Artificial Intelligence (AI) industry, focusing on the working conditions of Data Workers, particularly content moderators in Kenya. Despite AI's promise of automation, its development relies on intensive, low-paid, commonly outsourced human labour (Data Work: annotation, labelling, moderation) characterized by precarious working conditions, including exposure to traumatic content, low wages, isolation, unfair managerial practices, lack of regulation, and short-term contracts.

The research presents an exploratory case study of the African Content Moderators Union (ACMU), formed in Kenya in 2023. Driven by job insecurity, alleged union-busting, and psychological distress, the ACMU is a pioneering effort to unionize Data Workers providing services for global tech firms like Meta, OpenAI, and TikTok.

Based on semi-structured interviews with key members of the organization and secondary data, the study outlines the ACMU's motivations, demands, structure, and challenges. Despite its unofficial status, as it has struggled to register itself in Kenya, the Union has raised awareness of AI's "human cost," influenced research, supported workers, and raised important legal precedents against Big Tech.

The thesis concludes that worker-led initiatives like the ACMU are vital for fairness in the AI supply chain, contrasting AI hype with labour exploitation. It calls for policy interventions: recognizing Data Work, setting international labour standards, clarifying employment relationships, ensuring downstream accountability, and supporting collective bargaining rights.