Document details

Invisible Workers, Visible Harms: Perils and Precarities of AI Labour

Contains bibliogr. pp. 33-39

"Across South Asia, Southeast Asia, and Sub-Saharan Africa, data workers operate within fragmented and opaque supply chains characterised by informal employment arrangements, limited social protections, and a profound imbalance of power between workers, vendors, and the global technology firms that ultimately benefit from their labour. Many workers are engaged through short-term or micro-contracts, lack written terms of employment, and remain outside national labour law protections due to subcontracting structures designed to externalise risk. These conditions are compounded by intensive algorithmic management systems that monitor keystrokes, errors, and productivity in real time—practices that heighten stress, restrict autonomy, and render the labour process increasingly punitive.
For content moderators and workers handling sensitive data, the risks are particularly acute. Repeated exposure to violent, abusive, or graphic material contributes to severe mental health harms, including anxiety, depression, insomnia, and symptoms consistent with post-traumatic stress. Yet access to trauma-informed occupational health support remains extremely limited, and NDAs routinely prevent workers from reporting harms or seeking recourse. Despite these structural obstacles, workers are developing new forms of solidarity and resistance.
Drawing on desk research, secondary data, and stakeholder consultations across South Asia, Southeast Asia, and Sub-Saharan Africa, the report investigates the organisation, conditions and governance of this growing workforce. The report concludes with comprehensive recommendations spanning fair employment protections, occupational health standards specific to high-risk data work, transparency requirements for supply chains, mechanisms for collective worker voice, and binding international standards to prevent regulatory arbitrage." (Executive summary)
1 INTRODUCTION, 5
Data Work as the Foundation of AI -- Profit and Labour in the Global Labour Geography 7
2 LANDSCAPE | UNDERSTANDING THE WORKFORCE, 9
The Demographics of Data Workers -- Global South as AI’s Labour Periphery -- Mapping of Business Models
3 LABOUR CHALLENGES AND WORKING CONDITIONS OF DATA WORKERS, 14
Perils of Data Work and Content Moderation -- Psychological tolls -- Silencing of Data Workers -- The Human Cost of Automation -- Cross-regional findings from the Consultations and Interviews
4 EMERGING PATHWAYS FOR WORKER SOLIDARITY, 24
5 WAYS FORWARD: IMPROVING CONDITIONS FOR DATA WORKERS, 27
Recommendations for Supplier Countries (Global South) -- Recommendations for Sourcing Countries (Global North) -- Recommendations for international development cooperation
6 CONCLUSION, 32