When Does a Smart City Become a Surveillance City?

A camera used to count vehicles during a short traffic study and a camera used to identify people across public spaces may share the same hardware, but they create very different civic relationships. Calling both systems “smart” obscures the questions that matter: who is being observed, what is being inferred, how long records are retained, which agencies can access them, and what happens when a person cannot opt out. A smart-city system moves toward surveillance when data collection becomes persistent, identities become linkable, uses expand beyond the original purpose, and residents cannot meaningfully inspect or challenge the process. Accuracy alone does not resolve the issue. A perfectly accurate system can still be disproportionate, while an inaccurate one can create harm through false matches or selective enforcement. Cities need safeguards before procurement: a necessity and proportionality assessment, a public registry of sensing systems, strict retention and access policies, independent review, and clearly defined conditions for shutting systems down. Vendors should disclose capabilities that may be enabled later, because capability drift often occurs through software updates rather than the installation of new street hardware. The practical boundary is governance. A city remains accountable when it can explain why a system exists, demonstrate that less intrusive alternatives were considered, publish evidence of benefits, and stop collection when that evidence does not materialize. Without those conditions, urban intelligence becomes an imbalance of power rather than a public service.