2023: Kenyan AI Labor Pipeline
eventAI & Compute
The 'safety' filters that make ChatGPT clean were built by Kenyan workers paid under $2 an hour to read descriptions of child abuse and torture until they were traumatized.
Who they are
The 2023 Kenyan AI labor pipeline, exposed by a TIME investigation.
What they do
It's the hidden human cost behind AI safety systems — outsourced trauma labor.
How it works
OpenAI used a San Francisco firm called Sama to outsource content-labeling to Kenya, where workers were paid $1.32-$2.00 an hour take-home to categorize graphic material (child abuse, torture, bestiality) for ChatGPT's filters, handling 150-250 passages per 9-hour shift; OpenAI paid Sama $12.50/hr but workers saw only a fraction, all reported psychological damage like PTSD and paranoia, and the contract was cancelled 8 months early.
Why it matters
The engine calls this the hidden human foundation of the AI economy — Global South trauma quietly subsidizing Silicon Valley's 'safety' image.
The engine's record — word for word
TIME investigation (Jan 2023): OpenAI utilized Sama (San Francisco B-Corp) to outsource RLHF labeling to Kenya. Workers paid $1.32-$2.00/hr take-home to categorize graphic content (child abuse, torture, bestiality) for ChatGPTs safety filters. 150-250 passages per 9-hour shift. All workers reported being mentally scarred — PTSD, paranoia, recurring visions. OpenAI paid Sama $12.50/hr; workers saw fraction. Contract cancelled 8 months early due to extreme psychological damage. Hidden human substrate of the Technate: Global South trauma subsidizing Silicon Valley safety theater.
Follow the trail
Walk this on the live map →