The Global Color Line of AI Labor: Digital Colonialism in the OpenAI–Sama Supply Chain

Anis Rahman^1, @

1 : University of Washington

Despite the perception of AI as fully autonomous, human labor remains central to global AI production chain, although unevenly distributed (Irani, 2019). The hidden workforce, including data annotators, content moderators, infrastructure maintenance staff, and service employees, is dispersed globally and concentrated in the Global South (Casilli, 2025; Karanja, 2025). These undervalued “data janitors” perform essential yet precarious tasks such as collecting, labeling, and curating datasets for machine learning systems, often facing exploitation, job insecurity, and exposure to harmful content (Crawford, 2021; Irani, 2019; Muldoon et al., 2025). Critics contend that Western corporate AI creates new forms of extraction and dependency through a colonial supply chain that echoes historical imperial hierarchies and deepens global divisions of digital labor (Muldoon & Wu, 2023). This paper extends these discussions by arguing that the global supply chain of AI labor constitutes a form of geo-racial inequity between North and South, in which racialized workers from the Global South disproportionately bear the burdens of economic and psychological harms of exploitative data-cleaning work essential to AI development in the Global North. Drawing on W. E. B. Du Bois's (2007) concept of the “global color line,” the paper advances a Du Boisean critique of AI labor, contributing an intersectional lens (Collins, 2019) into the wider genealogy of digital colonialism thesis (Nothias, 2025). Using desk research and case study analysis, the paper examines OpenAI's labor contracts with Sama, a San Francisco-based firm employing workers in Kenya, Uganda, and India to label data for Silicon Valley clients including Alphabet, Meta, and Microsoft. Media reports revealed that Sama hired Kenyan workers to help make ChatGPT less toxic, exposing them to graphic content such as child sexual abuse, murder, and torture, while paying under two dollars per hour despite contractual rates of twelve dollars and fifty cents (Perrigo, 2023; Stahl et al., 2024). Sama later cancelled these contracts, though similar practices are pervasive in Africa's digital sweatshops (Muldoon et al., 2024). A critical media discourse analysis of investigative reports from both mainstream outlets (such as Time, The Guardian, and CBS News) and local alternative media highlights the racialized, classed, caste, and gendered dimensions of precarious data-cleaning labor in comparative context. These dimensions of human suffering, or at least the potential risks of labor exploitation, are often carefully concealed in recent finance-driven and promotional AI journalism produced within neoliberal mainstream media systems and industry reports, given the lure of billions of dollars in investment by U.S. hyperscalers and their local partners or competitors, as well as the lack of robust labor protection mechanisms. By foregrounding the suffering of data annotation workers, the paper contributes to an intersectional understanding of labor in global AI supply chains and calls for stronger labor protections, transparency, and ethical accountability in AI development. In conclusion, the paper suggests that an equity-concerned and intersectional analytical framework is essential for seeing through the AI hype (Bender & Hanna, 2025) while unveiling the matrix of oppression surrounding the global data center and its global color line, operating at scale.

Subject : : Abstract

Comment : Work cited: Bender, E. M., & Hanna, A. (2025). The AI con: How to fight big tech's hype and create the future we want. Random House. Casilli, A. (2025). Waiting for robots: The hidden hands of automation. University of Chicago. Collins, P. H. (2019). Intersectionality as critical social theory. Dule University Press. Crawford, K. (2021). The atlas of AI: Power, politics, and the planetary costs of artificial intelligence. Yale University Press. Du Bois, W.E.B. (2007). The world and Africa. Oxford University Press. Irani, L. (2019). Justice for Data Janitors. In S. Marcus & C. Zaloom (Ed.), Think in public: a public books reader (pp. 23-40). Columbia University Press. Karanja, N. (2025, March 27). Data from the South, AI in the North: An Uneven Distribution of Value. Strathmore University Center for Intellectual Property and Information Technology Law. https://cipit.org/data-from-the-south-ai-in-the-north-an-unevendistribution-of-value/ Muldoon, J., & Wu, B. A. (2023). Artificial Intelligence in the Colonial Matrix of Power. Philosophy & Technology, 36(80), 1-24. https://doi.org/10.1007/s13347-023-00687-8 Muldoon, J., Cant, C., Graham, M., & Ustek Spilda, F. (2025). The poverty of ethical AI: impact sourcing and AI supply chains. AI & society, 40, 529–543. https://doi.org/10.1007/s00146-023-01824-9 Muldoon, J., Graham, M., & Cant, C. (2024, July 6). Meet Mercy and Anita – the African workers driving the AI revolution, for just over a dollar an hour. The Guardian. https://www.theguardian.com/technology/article/2024/jul/06/mercy-anita-african-workers-ai-artificial-intelligence-exploitation-feeding-machine Nothias, T. (2025). An intellectual history of digital colonialism. Journal of Communication, 1–13, https://doi.org/10.1093/joc/jqaf003 Perrigo, B. (2023, January 18). Exclusive: OpenAI used Kenyan workers on less than $2 per hour to make ChatGPT less toxic. Time. https://time.com/6247678/openai-chatgpt-kenya-workers/ Stahl, L., Chasan, A., Bar-On, S., & Jung, J. (2024, November 24). Kenyan workers with AI jobs thought they had tickets to the future until the grim reality set in. CBS News. https://www.cbsnews.com/news/ai-work-kenya-exploitation-60-minutes/

Topics : Working conditions, occupational safety and health of workers in the human-in-the-loop

Keywords : Artificial Intelligence ; labor ; digital colonialism ; global color line ; intersectionality ; Open AI