This roundtable gathers academics, regulators, and practitioners to rethink how AI regulatory sandboxes can advance trustworthy innovation while reconfiguring accountability in data governance. Organised by the CybeRights Centre of the University of Florence, it aims to feature voices from the AI Office, EDPS, and national authorities to explore the promise and pitfalls of experimentation in EU regulatory practice. The discussion will deal with forward‑looking questions: Can sandboxes be more than compliance tools - acting instead as laboratories for rewriting AI and data governance? How might sandboxes redistribute trust and responsibility across public and private actors? What forms of evidence should (or should not) shape the next generation of digital regulation? Through dialogic exchange, participants will co‑develop insights for adaptive, ethically grounded models of regulatory experimentation. This will lay the groundwork for a joint publication documenting sandbox governance prototypes and their implications for trustworthy AI across Europe.