The Right Not to Be Modeled: Privacy, Surveillance, and Involuntary Model-Building

This essay reframes privacy as governance over modelability: the degree to which a person can be inferred from, classified, predicted, simulated, scored, or acted upon through available traces. It argues that privacy cannot be understood only as secrecy or control over direct disclosure, because AI-era systems increasingly act on derived patterns, probabilities, and classifications. The essay distinguishes visibility from modeling, sharing from surveillance, and disclosure from inference, then develops the concept of model-hardening: the process by which provisional or partial models become persistent, opaque, actionable, difficult to contest, and costly to revise. The argument is not anti-modeling; models are necessary for memory, coordination, care, safety, and administration. The concern is involuntary, consequential modeling in contexts where people cannot inspect, correct, refuse, or outgrow the models built from their traces. The essay proposes “the right not to be modeled” as a proportional governance principle: the more a model can affect access, treatment, reputation, opportunity, liberty, or future possibility, the more it should be bound by purpose, context, contestability, expiration, and limits on use.