Mophat Okinyi

At great personal cost, Mophat Okinyi helped to make ChatGPT a success. A former content moderator based in Kenya, his job was to read and label thousands of descriptions of toxic content—rape, child sexual abuse, sex with animals—to inform an algorithm that would help detect the kinds of things ChatGPT shouldn’t say. Okinyi says the work, for the outsourcing company Sama, was low-paid and grueling. He says he developed mental health problems and his wife left him.

Today, Okinyi is a new man. After blowing the whistle on the working conditions at his former employer, he helped to start the Content Moderators Union, the first union designed to protect the rights of AI data workers in Africa. Now the union’s chairperson, Okinyi has overseen its growth to more than 400 workers at several companies; worldwide, the total universe of AI data labellers is likely in the millions, by one estimate. He’s also navigating the organization through Kenya’s slow bureaucracy toward an official registration. 

OpenAI has said it was not aware of the working conditions at Sama, and no longer works with the contractor. A Sama spokesperson said in an email that “claims of mental health issues and/or family breakdowns were not brought to our attention during the time of employment,” and added that counseling was available to workers at all times.

With his meager savings, Okinyi this year separately founded an NGO, called Techworker Community Africa, which educates African AI workers and students about their rights, helps them avoid precarious short-term contracts, and advocates for better wages and mental-health support. “The biggest problem is that workers are not informed of their rights,” Okinyi says. “They can be easily exploited. That’s why we’re trying to give them some kind of training.”

*Disclosure: Investors in Sama include Salesforce, where TIME co-chair and owner Marc Benioff is CEO.

Buy your copy of the TIME100 AI issue here