"Worry about power concentration more than AI misalignment Vitalik Buterin identified two distinct categories of AI risk — multipolar (many bad actors with dangerous tools) and unipolar (one entity gaining irreversible dominance) — and argued that surveillance is the clearest current unipolar threat: "surveillance takes whoever is stronger and makes them even stronger — it removes spaces where pluralism can form, where counter-elites can coalesce." He grounded this in a concrete real-world observation about Russia, where improved living standards have coincided with facial-recognition-enabled suppression of protest. For any technologist or policymaker, this means the question "could this tool entrench the already powerful?" must be asked before "is this tool dangerous in the abstract?" Source: Vitalik Buterin, Tech Panel Discussion, EAC vs DACK"