Africa Digest News

Kenyan workers form a critical yet largely invisible backbone of the global AI industry.

Thousands of data annotators and content moderators in Nairobi and other Kenyan hubs label huge datasets, moderate harmful content, and refine outputs for leading AI models developed by companies such as OpenAI, Meta, and others.

Their labour powers the training of large language models and safety filters that enable generative AI systems used worldwide.

While this work drives technological advancement, recent media investigations and worker testimonies from 2025 and early 2026 highlight significant human costs, including psychological trauma and exploitative wage structures.

This discussion examines the dual realities faced by Kenyan workers: the documented challenges of their roles and the emerging initiatives aimed at transitioning them into higher-value positions within ethical AI development.

The Role of Kenyan Workers in Global AI Systems

Kenya has emerged as a major outsourcing hub for AI data work due to its youthful, English-speaking, tech-literate workforce and relatively low labour costs.

Workers perform essential tasks, such as:

• Annotating text, images, and videos to train machine-learning algorithms.
• Reviewing and classifying graphic, violent, or sexually explicit content for content moderation.
• Labelling data for applications ranging from chatbots to autonomous systems.

Much of this labour is outsourced through third-party firms (e.g., Sama) that contract with Big Tech. Kenyan workers thus contribute directly to the safety and accuracy of AI systems deployed globally, yet they remain at the edge of the value chain.

Reported Challenges: Psychological Toll and Wage Disparities

Media reports from 2025–2026 consistently document severe working conditions. Workers frequently encounter distressing material, including depictions of violence, child abuse, self-harm, and explicit content, for extended periods with inadequate mental-health support.

A 2025 Equidem survey of workers in Kenya (alongside Colombia and the Philippines) recorded numerous incidents of anxiety, depression, PTSD, and other psychological harms.

Over 140 former Kenyan content moderators involved in Meta-related cases have been diagnosed with severe PTSD, according to court filings and reports by The Guardian and Time.

Wage structures compound these issues. Kenyan data annotators and moderators typically earn between US$1 and US$2 per hour or as little as US$0.01 per task far below rates paid to workers in the United States or Europe performing equivalent work (US$10–US$25 per hour).

Pay disputes, withheld wages, opaque contracts, and strict non-disclosure agreements are common.

Investigations by 404 Media (March 2026), Black Agenda Report (March 2026), and Rest of World (December 2025) describe workers labouring up to 12 hours daily under algorithmic management, often without fixed salaries or benefits.

READ ALSO:How Kenya Is Powering AI Training Pipelines for OpenAI and Meta

Chinese AI firms have also increasingly tapped Kenyan students and graduates for video-labelling tasks at similarly low rates.

These conditions have prompted comparisons to a “new factory floor of exploitation,” as noted in a June 2025 Institute for Human Rights and Business analysis, where labour protections lag behind the rapid growth of AI demand.

Workers’ Collective Action and Policy Responses

In response, Kenyan workers have organised. The Data Labellers Association (DLA), launched in early 2025 and led by figures such as Secretary-General Michael Geoffrey Asia, advocates for fair pay, mental-health services, an end to exploitative NDAs, and improved benefits.

Broader efforts include the African Content Moderators Union and the Global Trade Union Alliance of Content Moderators.

Legal action has gained traction. Kenyan courts have heard landmark cases against Meta and contractors such as Sama, with a 2024 Court of Appeal ruling affirming that Meta can be sued in Kenya for labour practices.

Ongoing petitions and lawsuits address unfair dismissal, psychological harm, and modern slavery concerns.

Policy-level responses are also emerging: the Kenyan government is drafting regulations to address “digital colonialism” risks amid high youth unemployment (reported at 67% in mid-2025).

Kenya’s national AI Strategy 2025–2030 explicitly prioritises ethical AI governance, labour protections, and worker welfare.

Pathways to Empowerment: Skill-Building and Transition Programmes

Balanced against these challenges are deliberate efforts to upskill workers and move them beyond low-value annotation roles. Several initiatives target transitions into advanced machine-learning positions:

• Moringa School’s free AI upskilling programme (launched 2025 with Google funding) aims to train 3,600 Kenyan youth in practical AI competencies by December 2026, focusing on software engineers and expanding into education and healthcare.
• NextStep Foundation and Impact Outsourcing integrate data-annotation training with personal and professional development, emphasising fair wages and pathways for marginalised youth and women.
• The Open University of Kenya, in partnership with Pathways Technologies, offers professional programmes in AI fundamentals, machine learning, and related fields.
• Kenya’s national AI Strategy 2025–2030 places talent development at its core, promoting AI literacy, curriculum integration, and reskilling to create a workforce capable of higher-value contributions such as model development and ethical oversight.

These programmes seek to convert entry-level data work into stepping stones toward sustainable careers, reducing reliance on repetitive, high-trauma tasks.

Towards Ethical AI Development and Sustainable Labour Practices

The human cost of AI training underscores the need for systemic change. Ethical AI development requires not only technical safeguards but also fair labour standards across the global supply chain.

Stakeholders, including tech companies, governments, and civil society, must prioritise:

• Transparent contracts and living wages.
• Mandatory mental-health support and trauma-informed training.
• Clear accountability mechanisms that prevent outsourcing from shielding parent companies from responsibility.
• Investment in upskilling to ensure Kenyan workers benefit from, rather than merely support, the AI economy.

By embedding sustainable labour practices into AI governance, the industry can mitigate exploitation while harnessing Kenya’s talent for inclusive innovation.

Future Outlook

The human cost of training AI systems is borne disproportionately by workers in the Global South, including thousands of Kenyans exposed to distressing content and paid poverty wages.

Recent 2025–2026 media reports from 404 Media, Time, The Guardian, and others have amplified these realities and prompted worker organising through the Data Labellers Association and legal challenges in Kenyan courts.

Yet parallel progress in national AI strategy and targeted upskilling programmes offers a pathway toward more equitable participation.

Sustainable labour practices are not an optional add-on; they are foundational to responsible AI development.

Ronnie Paul is a seasoned writer and analyst with a prolific portfolio of over 1,000 published articles, specialising in fintech, cryptocurrency, climate change, and digital finance at Africa Digest News.