Safari AutoFill Security Issue Rears Its Head Once Again

Back in July, security researcher Jeremiah Grossman revealed a security issue that could allow malicious parties to take advantage of Safari's AutoFill feature to extract personal information from users' Address Book entries. At the time, Grossman reported that his report to Apple had gone essentially unacknowledged for nearly a month, but just six days later Apple released Safari 5.0.1 and 4.1.1 to address the problem.

Screenshot of Grossman's proof-of-concept test of new AutoFill exploit

To perform our attack requires tiny bit of end-user trickery. Two button presses to be precise. A malicious website detects (ie: IP address) the country the victim is from. For our purposes here we'll assume the "US." The attacker invisibly (CSS transparency) sets up the aforementioned form and forces the keystroke focus into the country element. Notice how this is done in the video on the right side of the screen, which only visible for demonstration purposes. Next the attacker entices the victim to type "U" (first character of "US") and then press "TAB." And BAM! That's it! Data stolen.

Grossman relates that he notified Apple of the newly-discovered exploit via email on August 10th and again a few days later. One week after that, he received a phone call from an Apple product security engineer with whom he had a "productive chat" about how the original vulnerability report from June had been handled, only to discover at the end of the conversation that the engineer had no idea that Grossman had reported the second issue a week and half prior.

As with the earlier exploit, users can protect themselves by simply turning off the AutoFill option to automatically populate forms with information from their Address Book cards. Grossman notes, however, that he is unsure how Apple plans to address the vulnerability while still maintaining the convenience of the AutoFill feature. While Apple's previous patch allowed Safari to automatically differentiate from the automated JavaScript-simulated keystrokes from real keystrokes, thus thwarting the original exploit, the new exploit relies on tricking the user into actually entering the necessary keystroke, a tactic that could be more difficult to address.