OpenID has warned of bugs in its authentication technology that create a possible means for hackers to modify data sent between sites.
The flaw is noteworthy because many high-profile sites – including Google, Yahoo! and Flickr – use the technology so that once users have logged into one site, they aren't constantly prompted for passwords. Thousands of smaller sites also use the technology.
The security weakness stems from an implementation flaw in authentication exchange, an extension to the OpenID system that gives sites the ability to exchange identity information between endpoints. The bug meant that proper checks on whether authentication information had been correctly signed were not carried out in some cases, thus creating a mechanism for hackers to offer false information that is accepted as genuine.
The security bug has been confirmed in OpenID4Java and Kay Framework, but is not necessarily limited to them. Both libraries have been updated. Janrain, Ping Identity and DotNetOpenAuth are immune from the bug.
There's no evidence that hackers have actually carried out any "psychic paper" trick along these lines, but the mere fact that the cryptography was weak is good reason enough for OpenID to advise sites to update their technology to a new version, without vulnerable implementations, or apply patches designed to plug the security hole, as explained in an advisory by OpenID here.
"For apps that are vulnerable, we recommend modifying application code to accept only signed attribute values as an initial step," it said.
OpenID credits security researchers Rui Wang, Shuo Chen and XiaoFeng Wang with discovering the flaw. ®