Poor Working Conditions for AI Data Annotators in Kenya and Colombia
Data annotators in Kenya and Colombia who label images and text to train generative AI systems reportedly earn as little as $0.01 per task while experiencing poor working conditions, including exposure to graphic content without psychological support and physical health impacts from long work hours.
What happened
Data annotators in Kenya and Colombia perform low-paid work labeling images and text to train generative AI systems. Kenyan worker Ivantous Kanyugi, 30, has classified and labeled images since 2018 for companies including Scale AI subsidiary Remotasks, earning only a few dollars or approximately $0.01 per task that may take hours.
Venezuelan worker Oscarinะฐ Fuentes, 35, working for five data annotation platforms in Colombia, earns between 5 and 25 cents per task. Workers report spending extended hours viewing graphic content from crime scenes without psychological support, experiencing visual problems, physical pain, and mental health issues from working up to 20 hours daily.
Reported impact
Data annotators experience visual problems, physical pain, and psychological distress from their work without adequate support. Workers describe their labor as invisible despite contributing to AI technology advancement.
The data annotation market was valued at $3.77 billion in 2024 and is projected to reach $17.1 billion by 2030, yet the workers who train these systems remain underpaid and unprotected.
Index analysis
Generative AI systems require human verification of training data and evaluation of AI-generated outputs. Most data annotators are aged 18โ30 from low-income countries, though the work is growing in the United States and Europe where wages are significantly higher.
Technology companies including Meta, Microsoft, and OpenAI outsource data annotation work to intermediary companies like Scale AI. Scale AI operates platforms including Outlier, which advertises specialist positions at $30โ50 per hour, while its Remotasks subsidiary in Kenya pays approximately $0.01 per task.
Governance controls
Kenya lacks legal regulations for the data annotation sector. The Data Labelers Association in Nairobi, with approximately 800 members, plans to announce a code of conduct for data annotation platforms to improve working conditions.
International labor organizations and civil society groups call for accountability from technology companies and fair employment contracts that include rest periods and psychological support. EU legislation reportedly does not address these workers.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.