Science Diplomacy and the Making of AI Governance: A Comparative Study of UNESCO’s Ethics of AI and the EU AI Act

This article investigates science diplomacy as concept and practice in the international governance of artificial intelligence. Although policy actors such as the European Commission and the Royal Society call for science diplomacy to inform global responses to AI, the concept remains weakly specified in AI governance debates. We address this gap through a comparative analysis of the EU’s Artificial Intelligence Act and UNESCO’s Recommendation on the Ethics of AI. Conceptually, the paper brings International Relations (IR) and Science and Technology Studies (STS) into conversation to develop a processual account of science diplomacy. From IR we draw on agreement making and institutional design to situate when expertise matters; from STS we mobilise co production to specify how authority is made in practice. Following Kaltofen and Acuto’s view of science diplomacy as a boundary problem, we treat each venue as a boundary organisation that assembles expertise, evidence, and authority in distinct institutional settings. Empirically, we conduct a mechanism focused documentary analysis of each process, comparing how the two organisations structure expert roles, curate evidence pathways, and navigate contested values through procedural tools such as assessments, audits, and reporting templates. Preliminary analysis indicates that both venues engage in knowledge brokerage, shaped by their differing constituencies, mandates, and institutional designs. These differences influence how each engages with stakeholders and sources views and evidence, including the EU’s closer engagement with private actors and UNESCO’s receipt of over 600 submissions from a broad range of actors. In turn, this shapes how each builds authority and legitimises its own expertise, as well as the expertise of others. We also trace how democracy and trust are envisioned and constructed, including the values each foregrounds, with UNESCO placing greater emphasis on human rights and the EU orienting more strongly toward the single market and its role as a regulator. Differences also emerge in how uncertainty is handled, for example, the EU’s regulatory approach tends to close down uncertainty through a stronger focus on risk, while UNESCO adopts a more forward looking stance and is currently inviting submissions from states in a more exploratory mode. These patterns have implications for both the study and practice of science diplomacy. They shed light on how science diplomacy operates in complex, fast moving technological contexts and underscore the role of institutional design in multilateral governance. The article brings Science Diplomacy, International Relations, and Science and Technology Studies into conversation by analysing how boundary organisations in AI governance assemble and legitimise expert authority amid epistemic and geopolitical asymmetries, and by examining how practices of coproduction shape contemporary digital governance.