AI could enable a tiny group of humans, or just one human, to take over a country and even the world. This risk is ballpark similar probability to AI takeover, and would be similarly bad if it happened. But there's been significantly (10X) less effort to reduce it. This talk aims to raise awareness and understanding of the risk, and discuss counter-measures.Tom is a Senior Research Analyst at Open Philanthropy He is currently working on assessing whether AI could lead to significant and long-lasting concentrations of power. Previously, he's worked on AI timelines (when AGI will be developed) and AI takeoff speeds (how rapidly AI capabilities will improve around the time of AGI). Prior to joining Open Philanthropy, Tom worked as a Data Scientist for education technology startup BridgeU and taught science at a UK comprehensive school. He has a Masters in Physics and Philosophy from the University of Oxford.