Web API Allows Phishing Attack

11 Security Sights Seen Only At Black Hat

A proof-of-concept phishing attack designed by a Stanford computer science graduate student offers a reminder that new Web technology can reinvigorate old risks.

Feross Aboukhadijeh, a Web designer, developer, and computer security researcher, has developed a way to spoof websites using the relatively new HTML5 Fullscreen API. As its name suggests, the API allows Web developers to present content in full-screen mode, without the distractions of Web browser interface elements.

Aboukhadijeh has published details of the attack on his website, which includes a link that opens a fake Bank of America page to demonstrate the technique. The attack uses the Fullscreen API to hide the interface elements of the user's browser--the chrome and the URL address bar--beneath an image of those interface elements. This prevents the user from knowing which website he is visiting.

Browser vendors have been aware of the potential for using this API for spoofing since it was first proposed. The Fullscreen API specification includes a "Security and Privacy Considerations" section that advises browser makers to implement an overlay element that tells the end user when full-screen mode has been initiated. "This is to prevent a site from spoofing the end user by recreating the user agent or even operating system environment when fullscreen," the specification states.

[ Read Why Huawei Has Congress Worried. ]

Unfortunately, Apple's Safari browser, version 6.01 and later, provides little or no sign that full-screen mode has been activated. Google Chrome, version 22 and later, offers some notice, though as Aboukhadijeh observes, the notification is "pretty subtle and easily missed." Mozilla Firefox, version 10 and later, alerts the user with a conspicuous notification.

Microsoft Internet Explorer 10 does not presently support the HTML5 Fullscreen API, although IE under Windows 8's Metro interface is always full screen. Microsoft confronted the issue of browser UI abuse in 2004 and placed constraints on chromeless popup windows in its Windows XP SP2 update to reduce the risk of spoofing.

Aboukhadijeh's attack depends on social engineering rather than flawed code. Although it's certainly less worrisome than a technical vulnerability that could allow the remote takeover of a computer, it's also less easily fixed--human gullibility can't be patched and defining adequate notification for an interface change isn't as simple as modifying code to prevent a crash.

There are a variety of ways to deceive people online and the only way to mitigate that risk is constant vigilance. In Apple's case, following the security recommendations in Fullscreen API specification would help, too.