Building Fair AI Supply Chains: Governance, Accountability, and the Fairwork Certification Scheme

Fairwork is an action research initiative that evaluates and scores companies in the digital economy according to the fairness of their working conditions. Grounded in five Fairwork Principles—pay, conditions, contracts, management, and representation—the project promotes higher labour standards through systematic benchmarking, transparency, and engagement with platform firms. Fairwork has assessed more than 800 companies worldwide and has facilitated measurable improvements in labour practices. Eighty-eight companies have implemented more than 440 pro-worker changes, including the provision of minimum or living wages, access to sickness insurance, anti-discrimination policies, and mechanisms for worker representation. 

This presentation examines the extension of the Fairwork methodology to the global artificial intelligence (AI) supply chain, focusing on the newly launched Fairwork Certification scheme. While AI production networks—encompassing data annotation, content moderation, and digital piecework—are often characterized by extreme opacity and fragmented subcontracting, the Fairwork Certification offers a rigorous evaluation and governance framework to bring these "hidden infrastructures" into alignment with international labour standards. Applying Fairwork's evidence-based framework to these often-invisible forms of labour does more than expose extractive processes; it provides a concrete mechanism for institutionalizing decent work. It is based on a globally renowned set of five core principles for fair working conditions —Fair Pay, Fair Conditions, Fair Contracts, Fair Management, and Fair Representation— and combines data from desk-research and interviews with workers, their representatives and platform management to serve as a bridge between corporate operations and the implementation of international labour standards as embodied in the International Labour Organization's (ILO) mandate for social justice. 

Crucially, this presentation highlights the political and governance-related features of the scheme. We argue that the Fairwork Certification functions as a non-state regulatory intervention that complements emerging mandatory human rights due diligence (mHRDD) legislations, such as the EU's Corporate Sustainability Due Diligence Directive (CS3D). By shifting from "checkbox compliance" to a four-phase evaluation model—involving discovery, action planning, implementation, and continuous monitoring—the scheme enables firms to move beyond ethical "hype" toward measurable accountability. 

Furthermore, the presentation emphasises the role of "social dialogue" and "impact sourcing" within the certification process. We demonstrate how the scheme facilitates democratic governance by requiring certified firms to recognize collective worker bodies and provide meaningful channels for grievance and appeal. 

By recent case studies of certified firms and the "pathways of change" triggered by our research-led interventions, we illustrate how the Fairwork Certification contributes to a human-centric AI supply chain. Ultimately, the project seeks to provide a tool to help transform global AI production from an unregulated "race to the bottom" into a transparent ecosystem grounded in the principles of social responsibility, collective bargaining, and the protection of human dignity.