I think this is an indication that app sandboxing is not good enough. It should be possible for the user to have control over everything necessary. The light that warns that the microphone is being used should also have functionality to disable it and to make the app "Request every time" or "Only when app is fullscreen" or similar. And if that is actually all happening, then sandboxing is working as necessary and therefore there is no actual trust issue, because the user doesn't have to trust the app anyway.
Adding to that, IMO there should also be an option that feeds bogus data to the app so it can't know if a resource is blocked. So many apps just refuse to work if you deny them the resources they don't need
That's a usability nightmare though. Imagine a user who clicked the wrong thing, then they complain that their mic is broken and plays analog TV hellscape sounds on conference calls.
Edit: Not saying I wouldn't use it myself, though.
Objective-See has a utility to do this called OverSight. It's a firewall for your webcams and mics. BUT do not mention this in the Zoom forums as it will be taken down.
