Fair Work Review

This document outlines initial correspondence by the Equiano Institute team regarding contributions for the Fair Work AI Kenya Review.

Artificial intelligence is often discussed in terms of models, investment, and innovation. Much less attention is paid to the people whose labour makes many AI systems possible. Through data annotation, content moderation, evaluation, testing, transcription, and other forms of digital work, workers in Kenya and across Africa contribute to AI systems used around the world, yet their experiences are often absent from conversations about AI governance.

The Fair Work for AI in Kenya project began with a simple question: how can journalism become a practical tool for AI accountability, and how can workers and communities become active participants in those conversations rather than simply subjects of them?

Rather than trying to answer every question about AI labour, this project sought to create spaces for discussion, learning, and collaboration. It explored how investigative reporting, public education, and civil society engagement can work together to make AI supply chains more visible and to encourage more worker-centred approaches to AI governance.

Outputs and Contributions

Although supported through a relatively small Pulitzer Center microgrant, Fair Work for AI in Kenya developed into more than a single workshop. The project created a growing body of educational resources, policy discussions, public engagements, and collaborative outputs centred on worker-led AI accountability. Rather than producing a single research report, the emphasis was on building practical resources that could continue to evolve beyond the grant period.

Fair Work AI Accountability Workshop

The project's central activity was the Fair Work AI Accountability Workshop, which brought together participants to explore AI through the interconnected lenses of labour, governance, accountability, and public-interest journalism.

Using reporting from the Pulitzer Center as an evidence base, participants examined how AI systems depend on human labour throughout their development and deployment. Rather than focusing exclusively on technical questions, the workshop encouraged participants to consider who performs this work, who bears the associated risks, and what responsibilities institutions should have toward workers and affected communities.

The workshop was designed as an interactive learning space, combining facilitated discussion, case studies, collaborative reflection, and practical exercises that participants could adapt within their own organisations and communities.

To extend access beyond those able to attend, the workshop was recorded and made publicly available on YouTube, allowing educators, researchers, and practitioners to continue engaging with the material after the event.

Fairwork Safeguards

One of the project's most significant practical outputs was the development of Fairwork Safeguards, an emerging framework that explores what responsible AI development should require beyond technical performance measures.

Rather than treating accountability as a question of compliance alone, the framework encourages organisations to consider worker wellbeing, transparency, institutional responsibility, participation, and mechanisms for redress throughout AI development and deployment.

The framework remains under active development but provides an initial structure for discussing worker-centred AI governance across academia, civil society, and industry.

Theory of Change

The project also produced an initial Theory of Change that captures the approach taken throughout the grant. The Theory of Change links investigative journalism, public education, facilitated dialogue, resource development, and policy engagement as mutually reinforcing activities.

This framework now serves as a guide for future work, helping connect individual workshops and educational activities to longer-term goals around worker participation and institutional accountability.

Educational Resources

Alongside the workshop itself, the project produced reusable educational materials designed to support future facilitators and educators.

These include workshop slides, discussion prompts, facilitation notes, background reading, and online resources published through the Fair Work for AI website. Rather than creating materials solely for a single event, the intention was to develop resources that other organisations could adapt to different educational and policy contexts.

This emphasis on open educational resources reflects the project's broader goal of strengthening public understanding of AI accountability through accessible civic education.

Redress & Accountability for Just AI

The conversations developed through the workshop also informed Redress & Accountability for Just AI, a discussion paper by Jonas Kgomo.

The paper explores what meaningful redress should look like when AI systems cause harm, arguing that accountability should extend beyond technical robustness to include institutional responsibility, accessible remedies, and recognition of the rights of workers and affected communities.

Rather than presenting redress as an afterthought, the paper positions it as a central component of trustworthy AI governance. Related themes were later taken up in reflections from the Just AI Conference, including the need for adaptive redress mechanisms once harms become observable in practice.

Language Input for Data & Society (LIDS)

The project also contributed to ongoing work on Language Input for Data & Society (LIDS): Licensing Data for Communities.

This initiative explores how communities might exercise greater agency over the data used to develop AI systems through more equitable approaches to licensing, governance, and participation.

Although distinct from the workshop itself, LIDS reflects many of the same underlying principles around fairness, accountability, and community representation that informed the Fair Work project.

London Data Week

The ideas explored during the Pulitzer Center project continued to develop through subsequent public engagement.

At London Data Week 2026, Sumona Bose facilitated a workshop moderated by Jonas Kgomo, bringing together participants from the United Kingdom, South Africa, Ghana, and Indonesia, including public health practitioners, psychologists, educators, lawyers, policy analysts, designers, researchers, and technologists.

The workshop examined artificial intelligence not simply as a technical innovation but as a social and economic infrastructure shaped by labour, governance, and power.

Participants reflected on hidden labour within AI supply chains, the underrepresentation of Global South communities in AI governance, regulatory challenges, and the emergence of what was described as agentic inequality—the growing disparity between those with meaningful access to capable AI systems and those without.

The discussion concluded with a shared reflection that fair AI requires accountability, dignity, care, transparency, participation, and practical mechanisms for redress.

Documenting these conversations through the London Data Week Updates helped extend the project's educational impact beyond the original workshop.

AI for Good Geneva

Members of the project also participated in discussions during AI for Good Geneva, where many of these themes were shared with international audiences working on AI governance and public policy.

While these engagements were not formal outputs of the grant itself, they provided opportunities to test ideas developed during the project, receive feedback from broader communities, and connect worker-centred accountability with wider international conversations about AI governance.

Building an Emerging Programme of Work

Looking across these activities, the most important outcome of the project may not be any single resource or event, but the emergence of a connected programme of work.

The Pulitzer Center microgrant provided an opportunity to experiment with how investigative journalism, workshops, educational materials, policy discussion, and public engagement can reinforce one another. Together, these activities contributed to an evolving body of work that includes practical workshops, governance frameworks, discussion papers, educational resources, and international dialogue.

While much remains to be done, the project demonstrated that relatively modest interventions can help connect journalism with civic education, strengthen conversations around worker-led AI accountability, and provide practical resources that others can continue developing.