We're assuming that all playtesting groups are playing under identical versions of the rules - but we've seen from varying developer blogs that many things, such as certain talents and the role and importance of magic items, are still being toyed with in several different iterations by the designers. What if not all outside groups were given the same versions of these rule segments, such that certain groups were playtesting paragon level play with minimal reliance on magic items, and other groups were playtesting mainly heroic play with a more standard level of magic? What if the rules effects of several talents varied somewhat from group to group? Even easier still, as we know the playtesting groups have pre-written characters, what if each group was given specific, different builds of the same classes?
This would be fairly difficult from a logistical standpoint, but with the details of the rules apparently still in somewhat of a flux from week to week, it isn't really that unlikely.
So, if a group were to decide to break their NDA by anonymously leaking information, they would have no way of knowing that they weren't the only group with access to that specific iteration of the rules. It would then be a fairly trivial manner for Wizards to examine the leaked data and determine which group that iteration was sent to, leaving them able to press charges for contract infringement.
This all might seem a bit too elaborate, but if you were a playtester considering the likelihood of having to duke it out with Hasbro's lawyers over leaking some of 4e's Warlock powers, you probably wouldn't completely dismiss the idea.