What you describe is an ideal Alice / Bob situation, setting just a single protected surface. In real world it's getting far more complex: what OS your PGP clie

What you describe is an ideal Alice / Bob situation, setting just a single protected surface. In real world it's getting far more complex: what OS your PGP client is running on? On what hardware? Is anything else running on the same device that may break out of its sandbox and inject the code in another process? And so on. > when Apple, in fact, explicitly allowed remote code execution by untrusted actors Can you please elaborate what do you mean by that?