> The app developer should not get to provide the message text for a prompt like that.
In App Store apps, where these messages get human review, it could more safely be allowed.
In App Store apps, where these messages get human review, it could more safely be allowed.
Shady developers would try very hard to get past such checks, for example by changing that message based on the date, IP address range, nearby WiFi networks, receipt of a payload, etc.
They could introduce an optional string-valued entitlement and check its text. That would be a lot easier but still require them to translate messages.
LLM should still be able to catch a misleading permission message though.