Security researcher Bogdan Calin identified the cross-site request forgery (CSRF) vulnerability after noticing that by default, all Apple devices are set to load remote images -- meaning images that haven't been sent with the email. "A malicious user can send you an email with an embedded 1x1 pixel image with the background color of your email client, so it is not visible," he said in a blog post. "The email client will load this image from a remote server."
Instead of setting the image to load from a remote server, however, an attacker could instead make the email perform an HTTP GET request that points to the URL for the router's administrative interface. Because numerous routers ship with default usernames and passwords, and many users fail to change those settings, the email could also authenticate to the administrative interface and alter the router's configuration, for example to change its DNS settings.
Thanks to the exploit, an attacker could change the router's DNS settings to point to an attacker-controlled server, enabling them to run a clickjacking scam -- redirecting users' search requests to sites of the attackers' own choosing -- as well as to eavesdrop on all Internet traffic flowing to or from the router.
[ Alleged Stratfor hacker's sentence may be more severe than a murderer's. Should LulzSec Suspect Face Life In Prison? ]
"To increase the chances of this attack succeeding, I can send multiple images in the email; one with the default username and password for the router and others with most common passwords," said Calin, who's a Web application security researcher at Acunetix. Using iFrames, the attack commands can also be executed in the correct order, for example first authenticating to the router, and then changing settings.
Calin said he's demonstrated the attack against two types of Asus routers (RT-N16 and RT-N56U), as well as the Arcor EasyBox A600 and routers from TP-Link. But many more types of routers are likely vulnerable. "Any router that accepts configuration changes from GET parameters and doesn't protect against CSRF should be vulnerable to this simple attack," he said. "I can also confirm that this attack works on iPhone, iPad and Mac's default mail client," but the attack could likely be extended to other operating systems and email clients.
This type of attack could also be adapted to target Gmail users. Although Gmail doesn't load remote images by default -- users must click a button to do so -- if remote images from a sender are downloaded once, Gmail will automatically download them from the same sender in the future. "So using such attack against a Gmail user is also very simple: send a plain text email that will intrigue the victim to reply and once he or she replies, send an email with a tracking image," Calin said.
Thankfully, the fix for this vulnerability is simple: Don't use default usernames on routers, and always set a strong, unique password on every router to make it less susceptible to brute-force password attacks.
Another option is to disable the automatic loading of remote images. For an iPhone or iPad, doing this requires navigating to the "Settings" menu, then selecting "Mail, Contacts, Calendars," and locating "Load Remote Images" in the mail settings. "You need to disable this option to protect yourself against this attack," said Calin. "However by disabling this option, if somebody sends you an email with an embedded image you will not be able to see it."