AI and Labor Law | Invisible Architects of Artificial Intelligence

Somewhere in Kenya, Joan Kinyua is sitting in bed with her computer. There's no room in her apartment for any other piece of furniture, let alone a desk. It's 12 noon. She's already prepared tea, water, and her lunch to avoid any reason for breaks. Two hours later, her official workday begins, lasting until about 2 or 3 a.m. She pays for her computer and internet access herself. "To do my job, I needed my own start-up capital," she recalls.

Kinyua worked for over five years as a data worker, the person who trains artificial intelligence (AI). She taught the software of self-driving cars to react to various aspects of road traffic, such as stop signs. She also trained the Roomba, the well-known robot vacuum cleaner. Data workers, says Kinyua, connected via a large screen at Re:publica , work in the so-called platform industry . This is the informal labor market in which temporary contracts are awarded flexibly and at short notice to job seekers, freelancers, or marginally employed people via online platforms. Being seen as part of this industry was a struggle, says Kinyua - for a long time, Uber drivers were considered a classic example.

The platform industry, as in Kinyua's case, often involves inhumane working hours, insufficient breaks, insecure contracts, and excessive pressure to perform. It seemed to her at the time that the more output she generated, the more she was demanded of, and the less she was ultimately paid. Shortly before Christmas, her company unexpectedly terminated her contract, and she suddenly found herself at home with her baby, unemployed. Becoming pregnant at her company felt like committing a crime, she recalls.

Today, Kinyua is president of the Data Labelers Association, a union for data workers in Kenya and Nairobi. The union members describe themselves as the "invisible architects behind AI." "We had to teach ourselves our labor rights. For example, that hiring without employment contracts is a violation of our rights," she says.

Joining her on the podium is Andreas Hänisch, deputy chairman of the German TikTok works council . The council has existed since 2022. He sees himself as "in a pretty good position" regarding his employment conditions. A significantly larger group of workers at TikTok, those who moderate content—that is, ensure that platform regulations are adhered to—work under "less favorable conditions." He says they are the reason Hänisch became part of the works council.

Many migrant workers work in this sector, and they are often unaware of their labor rights. They don't know, for example, that it's difficult to fire someone for unsatisfactory performance in Germany, Hänisch explains. Or that they aren't required to sign a mutual termination agreement. This applies especially to freelancers. Thus, technology companies in Germany are also exploiting the precarious situation of platform workers.

The increasingly collected information about workers could help change this. If workers take control of this data, it could open up new opportunities for them, including at the union level, writes data anthropologist Alexandra Mateescu. One example: Coworker.org, a program that helps workers understand how opaque wage models work.

According to Labor Minister Bärbel Bas (SPD), digitalization is the "future issue of our working society." After all, Germany has the third-highest robot density in the world. In the fall of 2024, during the "traffic light" coalition government, the ministry presented a draft bill on employee data protection to create greater legal certainty. At Re:publica, Bas announced that she would "revisit" this issue.

In data work itself, there are many moves toward complete automation. "However, human interaction and thus data workers will still be needed," Kinyua is convinced. Therefore, precarious jobs will remain – and there is potential for change. "It's never too late to establish a works council or join a union," Hänisch emphasizes.