Autopilot could kill the engine and lock the brakes without turning itself off. By turning itself off and giving control to a human mere instants before a crash it's effectively ensuring the engine isn't turned off (as there'll be no time to do it before the crash, and the human may not be in a fit state to do it afterwards).

But more to the point, regardless of whether there's a good reason for it to do that or not, it shouldn't be used to claim on a "technicality" that autopilot wasn't active "at the time of the crash", as clearly it meaningfully was at all points leading up to it.