So what happens when Alice Black-Hat exploits the government back door and copies compromising photos from hundreds or thousands of users' phones in order to blackmail those users? Okay, you say that users shouldn't have compromising photos on their phones? [Debatable, but for the sake of this argument I'll allow it.]
So let's consider a slightly different scenario. Alice uses the back door to intercept a text message from Bob's bank to Bob. That text message contains a security code and is intended to allow Bob to reset his password for access to his online banking accounts if he's forgotten it. This approach is a two factor system; Bob's social security number or security question answer is something he knows, and his phone is something he has. Alice quickly logs in, transfers money from Bob's account to one she can access from an ATM, grabs the money, and runs.
Who's responsible for replacing the funds Bob lost? Bob did nothing wrong. Neither did his bank; they received a valid request and sent the response to the owner of the account as agreed. The smart phone manufacturer? The government?