Mercy Mutemi, Executive Director of Oversight Lab Africa, has raised fresh concerns over what she describes as deep-rooted exploitation in Kenya’s growing AI labour sector, linking recent layoffs at Sama to long-standing structural problems in the industry.
Speaking on Radio Generation on Thursday, Mutemi said the challenges go beyond individual companies, pointing to a wider system that exposes workers to poor conditions, mental health risks, and weak protections, even as global demand for AI-related work continues to rise.
She explained that her organisation focuses on strategic litigation aimed at addressing “structural and systemic problems in how tech AI is structured in African countries,” with a strong emphasis on safeguarding workers in the digital economy.
Mutemi said the rapid shift toward advanced technologies has only deepened existing patterns of exploitation.
“The world has gone not just digital, but Uber digital. AI is the thing now, but we're seeing the pattern is exploitation from raw materials, data, labour,” she said, adding that the whole pattern from beginning to the end is exploitation.
Her remarks come amid renewed scrutiny of Sama, a company previously contracted to provide content moderation and data annotation services in Kenya, particularly over its links to Meta and recent job cuts.
Mutemi pointed to the dismissal of a large number of workers as a sign of deeper issues within the sector. “They released a notice that they are firing about 1000 of their workers, data labelers and data annotators,” she said, questioning the move at a time when such roles are in high demand globally.
“These are workers who train AI,” she added, noting that Kenya plays a key role in powering global AI systems through labour-intensive tasks such as data labelling and content moderation.
According to her, the concern is not Kenya’s participation in the sector, but how the work is organised. “We are witnessing the effects of poorly structured foundation when it comes to AI training work in Kenya,” she said, adding that workers are often exploited for short period of time, then get fired. A new thing comes. The same, same thing happens.”
Mutemi traced the issue back to 2022, when a whistleblower exposed the realities faced by content moderators. She said these workers were tasked with reviewing disturbing material to help train algorithms.
“A content moderator is essentially an AI training worker,” she said, describing how individuals spent long hours “looking at videos that you and me have come across on social media, pornography, people being decapitated, violence and self harm.”
She added that some workers were misled during recruitment, with one individual hired for an administrative role only to later find himself reviewing harmful content.
Once on the job, workers were required to handle large volumes of material under strict timelines. “You have to take it down within this number of seconds, because we need you to take down as many as possible as you train our algorithm,” she said.
?s=20Mutemi warned that such conditions can have serious mental health effects, worsened by low pay and unequal treatment.
She said Kenyan workers are often viewed as replaceable. “We can just go to slums and get more people,” she said, describing what she termed a “tiered system of torture.”
She also criticised recruitment approaches that target vulnerable communities. “When they say they are targeting people from poor backgrounds, what they really mean is that those individuals are so desperate they can be subjected to any working conditions,” she said.
The dispute has since escalated into a legal battle involving former moderators, Sama, and Meta, with workers seeking accountability over alleged rights violations.
Mutemi said attempts by workers to organise were met with retaliation. “Union busting happened, some of them were fired,” she said, adding that many were later blacklisted despite their specialised experience.
The case has progressed through the courts, including a key decision by the Court of Appeal on jurisdiction. “As long as the human rights issues involved, nobody has immunity, including an American company,” she said.
She argued that the issue goes beyond individual cases and raises broader concerns about the direction of technology. “How do we protect people in this system, where before there was tech, there was people, and you keep protecting them?” she asked.
As AI continues to expand, her remarks add to growing calls for stronger oversight, accountability, and a rethink of how technological growth is balanced with human rights.
Comments
Sign in with Google to comment, reply, and like comments.
Continue with Google