Computerized systems to monitor old people, perhaps partly demented, for their safety, work by means of AI-based surveillance, and their actions can feel oppressive to the people they "help".

However, the problem goes deeper: they contain proprietary software, so neither the people to be "served" nor the people who install the system have real control over it.

In particular, they can't be sure which sorts of events are being observed and reported to companies by the system, and used to profile or manipulate people.