LLMs Turn Surveillance Into Spying

Large language models and related AI now make it feasible to turn routine data (GPS pings, purchase records, search queries, photos, voice samples) into strong inferences about intent, health, and beliefs. That means state or corporate harvesting that once only tracked behavior can now be used to 'read' minds or predict dispositions, changing the moral and legal stakes of data collection. — If true, this shift requires new procurement limits, legal protections, and public debate because existing privacy law and norms treat collection and inference very differently.