Its possible to design internet connected systems that are secure and reliable. But usually "internet connected" means "dependent on continuous connectivity to some external service that needs to be actively maintained by a third party otherwise it'll stop working", which makes the "reliable" part difficult, and the fact that its exposed to a network with other devices on it and you have no way of knowing how competent that same third party is when it comes to security makes the "secure" part difficult.

Offline systems have way fewer failure modes such that they're often "secure and reliable" by just default, at least in comparison to the alternative.