Combined with the growth of online platforms, shifting patterns of information consumption, and new uses of personal data, AI has contributed to changes in how people interact online. The consequences of the enhanced personal tailoring – or targeting – that follows have already been visible in debates about ethical data use, and its consequences for public and political debate. Such debate is shaped by a variety of social and digital divides, which create power asymmetries in the digital landscape. These asymmetries affect whose interests are reflected in the development of AI, to whom the benefits and risks of deployment accrue, and what levers are available to direct technology development towards wider societal benefit. This trend highlights the importance of policy foundations that ensure trustworthy data governance and responsible AI research and innovation.