On Sunday 19 April 2026, Rotterdam's Kino hosted the Dutch premiere of In the Belly of AI, a documentary that travels from Finnish prisons to data labelling hubs in Bulgaria and Kenya to reveal what lies beneath the polished surface of artificial intelligence. The screening was organised by SocTalk and the Platform Labor Group, and drew an engaged audience of (ESSB) researchers, students, and professionals.
Directed by Henri Poulain and co-written by digital labor scholar Antonio Casilli and journalist Julien Goetz, the 2025 film makes visible the workers who annotate, clean, and categorise the data that feeds AI systems, often under precarious conditions, without knowing whose algorithms they are building or to what end. Alongside the human cost, the film documents the environmental footprint of the infrastructure required: vast data centres consuming land, water, and energy at a scale rarely acknowledged in mainstream narratives about technological progress.
A systemic problem, not a personal one
Following the screening, Casilli joined a panel of researchers including Francisca Grommé and Phuong Hoan Le from ESSB, alongside colleagues from Utrecht University, the University of Amsterdam, Erasmus School of History, Culture and Communication, and Erasmus School of Law. The conversation ranged across disciplines - EU law, sociology, communication science, digital geography - and converged on a shared concern: that existing regulatory frameworks, including the EU AI Act and the Platform Work Directive, remain insufficient to address the structural vulnerabilities of this workforce.
Panelists noted how data workers frequently operate at several removes from the companies that profit from their labour, a distance that obscures accountability and makes collective organisation difficult. The work itself, often described as a form of reproductive or care labour, can produce lasting psychological effects, including states of hypervigilance that extend beyond working hours into private life.
What can be done?
The evening closed with a question that proved difficult to set aside: how do we avoid being complicit in a system so deeply embedded in everyday life? The discussion moved deliberately away from individual responsibility as the primary frame, emphasising instead the need for institutional accountability, stronger regulation, and collective action. Some panelists pointed to emerging examples of more transparent AI development, such as EUR's own Desidera tool, designed to be explicit about how it operates and where its data comes from, as indicators of what a more socially responsible approach might look like.
Above all, the message was that change requires organised effort: supporting researchers, labor rights organisations, and political groups working toward fairer conditions and greater transparency in how AI systems are built and governed.
The event was organised by Sofie Schuller (UU | EUR), Anna Elias (ISS), Phuong Hoan Le (ESSB), and Charlotte Pennings (intern, Platform Labor Group).
