![\"\"](\"https:\/\/www.europesays.com\/uk\/wp-content\/uploads\/2026\/08\/ai-tag-comp-0679.gif\")
\n\t\tData Labelling is a job you probably haven\u2019t heard of, two data labellers told Metro (Picture: Getty\/Metro<\/p>\n
\t\t
\n\t\tData Labelling is a job you probably haven\u2019t heard of, two data labellers told Metro (Picture: Getty\/Metro<\/p>\n
Joan Kinyua, 38, has thought a lot about how she\u2019d kill, skin, cook and eat a human, all while her two-year-old girl played in the next room.<\/p>\n
But for Joan, who lives in Nairobi, Kenya<\/a>, this was once all part of her job.<\/p>\n She\u2019s a former data labeller, the humans behind AI<\/a>.<\/p>\n \u2018People are really strange,\u2019 Joan, the co-founder of the Data Labelers Association (DLA)<\/a>, which has nearly 900 members worldwide fighting for better pay and perks, tells Metro<\/strong>.<\/p>\n \u2018These are things that I thought nobody ever thought about, especially publicly asking on laptops.<\/p>\n \u2018When I learnt this was all helping tech bros to make sure people don\u2019t use AI to harm others\u2026 I couldn\u2019t touch a laptop or social media<\/a> again.\u2019<\/p>\n What is a data labeller?<\/p>\n \t\t AI doesn\u2019t come ready-made knowing the difference between a tree and a car<\/a>, or why you shouldn\u2019t tell someone how to cook a human.<\/p>\n That\u2019s where data labellers come in. Also called data annotators or AI tutors, they \u2018tag\u2019 text, images and videos to teach algorithms.<\/p>\n They, for example, annotate photos of roads \u2013 that smudge is a child, that red circle a traffic light \u2013 so robotic cars<\/a> can drive. Or, as Joan did, label gruesome surgical videos so AI can diagnose health issues<\/a>. <\/p>\n The same work keeps AI safe to use<\/a>.<\/p>\n \u2018A cannibal cannot ask what human flesh tastes like, or prompt how to skin a person alive,\u2019 Joan says.<\/p>\n \u2018Now you can\u2019t ask stuff like that. It\u2019s not because they taught themselves, but somebody somewhere sat down and thought of the weirdest, most gruesome, most horrific things a person could ask.\u2019<\/p>\n \t\t Rather than cannibalism, former data worker Michael Geoffrey Asia tagged pornography for nearly a year<\/a> so it\u2019s searchable: What do the actors look like? What position are they in? Among other things. <\/p>\n This left the secretary general of the DLA with insomnia, PTSD and trouble having sex.<\/p>\n \u2018Having to witness three huge guys molest a 13-year-old, that remained with me,\u2019 Michael says.<\/p>\n \u2018I have a daughter. It put me in a position where I couldn\u2019t trust any other man close to my child. A fear that you cannot fight, even from within.\u2019<\/p>\n Some data labellers don\u2019t know their employers as they\u2019re posted to job boards via a \u2018middle man\u2019, allowing companies to keep work private from snooping competitors.<\/p>\n Michael even signed non-disclosure agreements, preventing him from discussing his work.<\/p>\n \t\t With few ways to get support, Michael finally received therapy through a medical centre as he was caring for a cancer<\/a> patient.<\/p>\n The author of The Emotional Labor Behind AI Intimacy<\/a>\u2019s next gig was \u2018training\u2019 sexbots.<\/p>\n \u2018There is a man on the other side trying to seduce me<\/a>, believing I\u2019m a woman,\u2019 he says. \u2018Or I\u2019m supposed to act gay, or work with lesbians.<\/p>\n \u2018The morning after, you go to church. You\u2019re praying, asking, \u201cDid I do the right thing? Am I pretending?\u201d\u2018<\/p>\n Joan also \u2018trained\u2019 companion bots, which create lifelike friends and partners<\/a>, but she believes she was chatting with real people.<\/p>\n \u2018Somebody was 100% in love with it,\u2019 Joan says. \u2018I viewed the human as vulnerable; they opened up about how they\u2019re struggling to find love, telling you their deepest secrets.\u2019<\/p>\n How do they get jobs?<\/p>\n Metro <\/strong>was shown a job board filled with dozens of data labelling positions. <\/p>\n \u2018Inspector Genital \ud83d\udd75\ufe0f,\u2019 read one listing. \u2018In this job, you will select Real Genitals, Animated Genitals or No, depending on if the image contains any genitals.\u2019<\/p>\n Do that 1,000 times and you\u2019ll earn $0.30<\/p>\n \t\t \t\t Other gigs included:<\/p>\n Many asked the user to complete a task within a minute, Metro <\/strong>saw. <\/p>\n \u2018Even if you\u2019re given, \u201chow do you molest a two-year-old,\u201d despite having a two-year-old in the house, the task is timed,\u2019 Joan says.<\/p>\n \u2018AI labellers are refused breaks\u2019<\/p>\n The Data Workers\u2019 Enquiry<\/a>, representing workers from nine countries, has found <\/a>that AI tutoring can increase mental health issues and drug dependency.<\/p>\n Member Camilla Salim Wagner isn\u2019t surprised by Joan and Michael\u2019s experiences. AI is seen as Silicon Valley magic<\/a>, meaning it\u2019s easy to forget the humans behind it, he says.<\/p>\n \u2018If it were widely known that workers who make AI systems are routinely refused breaks during the day, forced to work uncompensated overtime and exposed to such a level of toxic or disturbing content that they carry psychological scars for years, all while still having to worry if the next paycheck will cover their bills<\/a>, this would significantly harm the reputation and credibility of AI companies,\u2019 Wagner adds.<\/p>\n<\/p>\n The research intern at the Distributed AI Research Institute says tech giants look to African nations for data labelling to cut costs.<\/p>\n The continent has scores of highly-skilled tech workers, and while many countries have strict labour laws, not all do<\/a>.<\/p>\n The data collection and labelling market was valued at about \u00a316billion <\/a>last year and is projected to balloon to \u00a388billion in 2034.<\/p>\n \u2018We\u2019re just being used to make billionaires and trillionaires,\u2019 says Joan.<\/p>\n Still, Michael feels that the work data labellers do is human \u2013 even if the result is for something that is anything but.<\/p>\n \u2018Everyone is so determined to train machines how to respond with empathy,\u2019 he adds.<\/p>\n \u2018But no one has ever trained people to love people like us, who have been working on this information all along.\u2019<\/p>\n Get in touch with our news team by emailing us at webnews@metro.co.uk<\/a>.<\/strong><\/strong><\/strong><\/p>\n For more stories like this, <\/strong>check our news page<\/strong><\/a>.<\/p>\n Arrow Arrow Arrow![\"\"](\"https:\/\/www.europesays.com\/uk\/wp-content\/uploads\/2026\/08\/comp-1785928213.png\")
\n\t\tJoan Kinyua (left) and Michael Geoffrey Asia are former data workers (Picture: Wanja Ngina\/Michael Geoffrey Asia)<\/p>\n![\"Tallinn,](\"https:\/\/www.europesays.com\/uk\/wp-content\/uploads\/2026\/08\/SEI_279438752-ff83.jpg\")
\n\t\tData labelling helps AI identify patterns (Picture: Shutterstock\/M-Production)<\/p>\n![\"close](\"https:\/\/www.europesays.com\/uk\/wp-content\/uploads\/2026\/08\/SEI_260072714-af0e.jpg\")
\n\t\tData labellers may have to watch porn frame by frame (Picture: Shutterstock\/Funstock)<\/p>\n![\"\"](\"https:\/\/www.europesays.com\/uk\/wp-content\/uploads\/2026\/08\/SEI_305887937-9af1-e1785922096117.jpg\")
\n\t\tAll jobs are timed<\/p>\n![\"'I](\"https:\/\/www.europesays.com\/uk\/wp-content\/uploads\/2026\/08\/SEI_305887938-52e6.jpg\")
\n\t\tSome of the images included nudes, detailed fan art and stock images<\/p>\n\n
\nMORE: People say they hate AI slop \u2014 but do you really?<\/a><\/p>\n
\nMORE: I was convinced I was going to meet my soulmate \u2013 but I had been lied to in the worst way<\/a><\/p>\n
\nMORE: Google U-turns on AI Earth image feature after users \u2018collapse Eiffel Tower\u2019<\/a><\/p>\n