Back to the MIT repository
6. Socioeconomic and Environmental1 - Pre-deployment

Human exploitation

When workers who train AI models such as ghost workers are not provided with adequate working conditions, fair compensation, and good health care benefits that also include mental health.

Source: MIT AI Risk Repositorymit1334

ENTITY

1 - Human

INTENT

3 - Other

TIMING

1 - Pre-deployment

Risk ID

mit1334

Domain lineage

6. Socioeconomic and Environmental

262 mapped risks

6.2 > Increased inequality and decline in employment quality

Mitigation strategy

1. Implement a mandate for **Responsible Value-Chain Contracting** that requires prime contractors and AI developers to establish and enforce living-income benchmarks, fair rates, predictable work volumes, and realistic deadlines across all third-party vendors and BPOs. Contracts must embed shared responsibility for worker outcomes and provide funding for these essential protections. 2. Establish and fund **Comprehensive Occupational Health and Safety Protocols**, with specific emphasis on psychosocial risk mitigation. This includes mandatory, evidence-based mental health support, trauma-informed care, and adequate protective safeguards (e.g., protected breaks, task rotation, content filters) for all workers exposed to harmful or graphic content. 3. Guarantee and protect **Labor Rights and Collective Agency** for all data workers, irrespective of their employment classification (e.g., contractor or employee). Policymakers and organizations must ensure the right to organize, collectively bargain, and access effective grievance mechanisms without fear of retaliation, thereby establishing a formal mechanism for improving working conditions and pay.