We've known for literally decades that that doesn't actually work, for several reasons:

1. People are conditioned to ignore warnings. There are way too many benign warnings in the world; you can't read them all.

2. Even when people wouldn't ignore them, in cases where they are being tricked by scammers it's easy for the scammer to talk people into accepting them.

3. Those sorts of warnings aren't actionable. You're installing a new app. It appears legit. You want to use it. You get a warning like "this app hasn't been verified; it might be malware!". What can you do with the information? Absolutely nothing. 99.9999% of users have zero way of doing any deeper check to see whether it actually is malware. Their only options are to give up and go home, or just hope that the warning is wrong. Even I - a highly technical user - get zero value from things like Windows' smart screen. "The app you're running hasn't been signed! It might be malware!". Err yeah sure. I'm not going to reverse engineer it to check am I?

I think their solution of allowing you to disable the restriction with a one-time one-day delay is actually a really reasonable solution. As long as they don't go further than that - the risk is that it is just a temporary placation and they'll ditch that option in a few years.

The problem is easy to solve by making 99% of all apps normal apps that don't get any special privileges and don't require any developer certification, and having a certified developer program with heavily locked down run mode for the 1% of high security apps like banking and payment apps. It's not hard to attest unambiguously to the user in some way whether they are running one of these rare secure apps or a normal one, a restricted API suffices but you could also just add an LED for it.

You can't possibly convince me that Google couldn't develop something like that if they wanted to.