Artificial intelligence is often framed as a technical system plagued by “bias,” but the roots of bias run much deeper. They emerge not only from skewed datasets or flawed algorithms, but also from political choices about what knowledge counts, who is represented, and whose interests are served. This session explores the intersections of AI bias, politics, and research governance. Drawing on cases from healthcare and beyond, it examines how institutional structures—ethics boards, funding bodies, and regulatory regimes—shape the way bias is recognized, ignored, or reproduced in AI systems. The session will argue that bias is less a computational glitch and more a reflection of entrenched power relations, making it a bioethical issue of urgent importance. Participants will be invited to rethink research governance for the AI era, considering transparency, contextual fairness, and participatory ethics as pillars for more trustworthy and equitable innovation.