Mophat Okinyi often experiences vivid images when he's alone or about to sleep. Okinyi, a former content moderator for Open AI's ChatGPT in Nairobi, Kenya, is among four individuals in the same position who have submitted a petition to the Kenyan government. Their petition calls for an investigation into what they claim are exploitative conditions endured by contractors who review the content that powers artificial intelligence programs. According to a study, ChatGPT demonstrates better performance than undergraduates in solving SAT problems. Okinyi describes the toll it took on his mental health, stating that he would go through up to 700 text passages daily, many of which depicted graphic sexual violence. He found himself avoiding people after reading texts about rapists and projecting paranoid narratives onto those around him. Eventually, his wife noticed the change in him and decided to leave, leaving him devastated as she was pregnant at the time. The petition filed by the moderators is directly related to the contract between OpenAI and Sama, a California-based data annotation services company that employs content moderators worldwide. The content moderators in Nairobi allege that while working for Sama, they experienced psychological trauma, low pay, and abrupt dismissals. The moderators claim that as part of their tasks, they were exposed to texts and images depicting violent scenes such as self-harm, murder, rape, necrophilia, child abuse, bestiality, and incest. They argue that they were not adequately informed of the extreme nature of the content they would be reviewing and were not provided with sufficient psychological support. According to a spokesperson from Sama, the workers were paid between $1. 46 and $3. 74 per hour. When the contract with OpenAI was terminated prematurely, the moderators were left without income while dealing with the trauma. The petitioners share their personal experiences, including Richard Mathenge, who felt abandoned after the contract ended and suffered from serious trauma. Alex Kairu, another petitioner, was offered a different role by Sama after the contract termination but witnessed his mental health deteriorate. The moderators believe that their experiences were overlooked, and they emphasize the need for better support and understanding from their employers. OpenAI declined to comment on the matter, while Sama stated that the moderators had access to mental health therapists and medical benefits. They also asserted their commitment to fair employment practices and complying with applicable legislation. The rise of generative AI, represented by ChatGPT, has raised concerns about the potential impact on various industries.

However, the focus has often been on the human labor behind AI's success. The demand for data labeling and collection services has led to an industry that is expected to be worth over $14 billion by 2030. Much of this work is being done in countries like Kenya, India, and the Philippines, where a pool of multilingual workers is available at a fraction of the cost. Nairobi has become a global hotspot for this type of work due to the economic conditions, high English proficiency, and diverse workforce. Sama was able to recruit young, educated Kenyans who were eager for employment. However, as the moderators' experiences reveal, what initially seemed like an ideal job turned into a traumatic and monotonous task. The moderators describe reading texts that depict horrifying scenarios, leading to emotional distress and a decline in their mental well-being. The petitioners claim that Sama provided inadequate psychological support, although the company disputes this. Sama's website emphasizes its commitment to improving employment and income outcomes. The company has major clients such as Google, Microsoft, and Ebay. However, in 2022, Sama terminated its partnership with OpenAI, citing a focus on computer vision data annotation solutions. Former ChatGPT moderators are now calling for new legislation to regulate the outsourcing of "harmful and dangerous technology work" in Kenya. They also demand that existing laws recognize exposure to harmful content as an occupational hazard. The moderators want an investigation into the ministry of labor's failure to protect Kenyan youth from outsourcing companies. The Kenyan government and companies like OpenAI share a significant responsibility in addressing these issues. Cori Crider, director of Foxglove, a non-profit legal NGO supporting the case, believes that tech companies must acknowledge their role as true employers, despite outsourcing this work. The formation of the first trade union for content moderators in Africa and recent legal rulings highlight the growing attention on the rights of outsourced workers. Moving forward, Okinyi finds solace in the thought that he has protected ChatGPT's users, likening himself to a soldier who takes bullets for the greater good. Despite the lasting impact on his well-being, he considers himself a hero.