My eye was caught by a report in Biotechniques news that the NIH had removed access to two genetic databases following research that indicated that you could extract an individual's data from a set of pooled data.
I had not been able to work out exactly which databases - see below for something I have just read that might shed light.
**Later note** - this story is discussed in Science, whose report suggests that it is not whole databases that have been pulled, but some data from within particular databases. Some data has been pulled by the NIH from dbGAP, and from a cancer genetics database called CGEMS (http://cgems.cancer.gov/, which is currently displaying a news item saying that some data is temporarily unavailable for public posting). The Wellcome Trust has pulled data too - the Science report does not give details.
The story is also reported in the LA Times, Peter Suber's Open Access News, and (mentioned by Suber) Nature News. The research itself is in PLoS Genetics.