Thinking Upstream: Ethics and Policy Opportunities in AI Supply Chains

AI production involves hidden labor exploitation in supply chains: Just as Nike faced backlash for poor factory conditions, AI companies like OpenAI have outsourced content moderation work to subcontractors, resulting in workers being paid less than $2 per hour to filter traumatic content while facing inadequate mental health support—demonstrating that ethical concerns in AI extend beyond the final product to upstream production processes.

Corporate responsibility requires direct accountability, not outsourcing ethics: The paper argues that companies cannot simply disclaim responsibility by using subcontractors, drawing a parallel to Nike's initial denial of responsibility before public pressure forced them to take ownership of their supply chain—suggesting AI companies must similarly be held accountable for conditions throughout their production pipeline.

Policy interventions must address the "upstream" of AI development: Rather than focusing only on AI model outputs and safety, the paper positions AI ethics as fundamentally a supply chain problem, indicating that policymakers should examine and regulate the working conditions, compensation, and welfare of workers involved in AI data preparation and content moderation.

Current corporate structures enable ethical evasion: The paper highlights how outsourcing arrangements allow AI companies to distance themselves from worker welfare issues, suggesting that policy solutions must address the organizational and contractual structures that enable companies to avoid direct responsibility for their supply chain labor practices.