Here's the thing about the government telling you an AI system is "not for surveillance": nobody says that about a traffic light. An official had to come out this week and specifically deny that Hong Kong's new City Brain AI is watching people, because enough people asked. That's not nothing. City Brain is the smart-city platform meant to manage traffic, energy, emergency response, all the boring plumbing of a modern city. Grand ambitions. Data everywhere. And the first question everyone had wasn't "will it work," it was "is it watching me."

Look, maybe it really is just traffic lights and bin collection. But you don't issue a press statement promising you're not doing the thing unless people already think you might be doing the thing. Same week, we also got Chief Secretary Eric Chan (via Paul Lam's ministry, warning on AI misuse) telling everyone to be careful about AI "breakouts" and bad actors gaming these systems. So the message from Central this week is: trust the algorithm, but also, be scared of the algorithm. Pick one, guys. The honest version is simpler: a system that touches this much city data will always need more than a denial to earn trust. A statement isn't an audit.