Rethinking Data Sovereignty through Convivial AI

Artificial intelligence is no longer confined to specialized domains: it increasingly underpins the very infrastructures of everyday life. Yet, as its presence grows, long-standing frameworks for data sovereignty are showing their limitations, particularly when it comes to enabling democratic oversight. In this chapter, the authors suggest moving away from narrowly defensive ideas of data control and toward a more generative, participatory understanding of digital agency. This shift is explored through the concept of Convivial AI, which treats AI not as an external system acting on society, but as a set of tools situated within communities and accountable to them. The argument unfolds through three pilot projects carried out in Castellammare di Stabia and Atena Lucana (Italy), and Utrecht (the Netherlands). These experiments combined digital and analogue tools in a hybrid methodology designed to support the collaborative creation, interpretation, and governance of data. The results are mixed, but telling. On the one hand, the approach fostered ethical co-design, inclusion, and mutual learning across diverse participants. On the other, it ran up against familiar barriers: institutional rigidity, uneven data literacy, and the challenges of sustaining engagement over time. Rather than offering a fixed model, the chapter contributes to ongoing debates about digital sovereignty by putting forward Convivial AI as a situated and evolving practice. It frames data as a commons, something to be cared for collectively. In doing so, it invites communities to reclaim agency in shaping technological futures aligned with democratic values and shared priorities.