I'm reading John Battelle's book, The Search, and the recent release of search data by AOL couldn't be a better example for his metaphor of the Database of Intentions. What comes out of this giant screw-up is as fascinating as AOL mistake is big.
It's pretty obvious that with such amount of data, some of it can easily lead to personally identifiable information. As note several commenters to the aforementioned post on Google Blogoscoped, using the timestamp information will allow some sites owners to link 3 months of a user's search track to an IP, or directly a name if they have a registration record or profile there. It shows a lot about how Americans regard privacy issues. See also posts by Ars Technica, TechCrunch, Valleywag citing the NYT which took no time to identify one guy out of this so-called "anonymous" data, and John Battelle. There are probably a lot more stories already published or in the waiting, since the released data is now floating around the internet for a long time, although AOL removed it promptly after the uproar.
It also gives a lot of hints on how this gigantic amount of individual intentions can be exploited, and will be exploited no matter what, for this is a goldmine for both commercial and political interests.