The day that Adobe announced Flash CS5 with iPhone support, I took apart several of the Flash-generated iPhone applications and tweeted my findings. Louis Gerbarg has done a more detailed dissection of the Flash-generated iPhone ipa files (the actual file that gets downloaded from the App Store) than I did. He found the use of multiple (about a dozen) private API calls being used after doing what he calls a "cursory examination". His opinion of the size of the applications jived with mine as well (unnecessarily bloated), and he throws in some more concrete evidence showing that these apps do, in fact, perform poorly on iPhones prior to the 3Gs.
Louis' findings would cast some doubt on Adobe's claim that they didn't do any reverse engineering. Of course, in my mind, there already was a fair amount of doubt on that point.