Nature Machine Intelligence

Sharp, M., Bilgin, O., Gabriel, I. et al. Agentic inequality. Nat Mach Intell (2026). https://doi.org/10.1038/s42256-026-01318-5

As autonomous AI agents capable of complex planning and action are deployed more widely, differences in who can access them, how capable they are and how many of them users can deploy are likely to shape the distribution of power and outcomes at a societal level. We characterize these disparities as agentic inequality. We argue that agents could either deepen existing divides or, under the right conditions, mitigate them. The Perspective makes three contributions. First, it develops a framework for analysing agentic inequality across three dimensions—availability, quality and quantity. Second, it argues that agentic inequality is not the same as earlier technological divides because AI agents function as autonomous delegates (rather than tools) and so generate new asymmetries through scalable goal delegation and direct agent-to-agent competition. Third, it analyses the technical, socioeconomic and political drivers of agentic inequality, such as model release strategies and market incentives, that are likely to shape the distribution of agentic power. We conclude by setting out a research agenda for governance.