Okay, I'm going to try to redirect this thread a bit from the long, drawn out discussion about legal requirements for model releases of explicit images (and the related record keeping), because I think that is only one small aspect of issues.
I agree with those who say there is a low risk of people accidentally finding images of an explicit nature in Wikipedia articles that are not directly related to those subjects. I do agree that at least some Wikipedia projects seem to have a disproportionately large collection of such articles, and that some of them are poorly named or identified, so that someone looking up a term that is used both in relation to a non-sexual topic and a sexual topic may get a bit of a surprise, and that needs to be addressed.
On the Commons side of things, I think there has been an over-aggressive campaign to extract "license compliant" images from Flickr and other non-WMF repositories that include subjects who were very unlikely to know that their image was going to be made available on Commons. I believe that whoever uploads those images to Commons has a personal responsibility to verify that all of the subjects in those images was aware of, and agrees to, the licensing terms. I also believe that it should become part of the process that prior to uploading such images, the person uploading to Commons confirms with the Flickr uploader that the terms of the license are correct, and that there are suitable model releases where applicable.
Let's not worry so much about what courts have decided, and pay more attention to developing best practices within our own projects.
Risker/Anne