Reading blogs could cause your computer to catch a virtual cold, said a leading security expert Wednesday at the Black Hat USA conference.
Internet users who employ Web-based services such as Bloglines or Web browsers such as Firefox to read Web site feeds and blogs are vulnerable to embedded malicious code that can install spyware, log users’ passwords, scan PCs and corporate networks for open ports and more, said Caleb Sima, chief technology officer at SPI Dynamics Inc., an Atlanta-based Web application security company.
So far, only a few proof-of-concept attacks against blog readers from Google and Yahoo have occurred, Sima said, though he believes that more are on the way.
“The only reason we haven’t had a lot of problems yet is because no one has really thought of it,” he said. According to Sima, software and services used to download feeds transmitted via the RSS or Atom formats can unwittingly download and execute JavaScript code buried within the text.
JavaScript is a scripting programming language that has become popular as a way of providing rich interactivity to Web sites. But it can also be employed by hackers to take over users’ computers, sniff for vulnerabilities and more.
“The possibilities are limited only by how creative you are,” Sima said, who calls the technique “feed injection.” Sima said this type of attack is a variant of cross-site scripting, a popular way of attacking Web sites by entering HTML or JavaScript commands instead of text, which can cause errors and disable the site, leaving it open to hackers.
Seemingly random strings of characters such as “<" are often converted by Web sites or blog readers into the "<" character. That signifies a left tag, which tells Web sites and software to treat any text between it and a right tag character as executable code.
This way, seemingly garbled text can hide malicious JavaScript commands that can do damage without having to install or run an outside file, a telltale sign that would normally alert an antivirus or antispyware program.
But don’t users sign up for RSS feeds only from Web publishers they already know and trust? True, said Sima, but there are many ways those publishers could be compromised. A Web feed could contain a link to another Web site or blog that’s hosting malicious JavaScript. Or the Web feed’s author could unknowingly paste that JavaScript into his own blog. Or a blog may have an area allowing readers to post public comments. Those can also store malicious bits of JavaScript, Sima said.
Finally, because RSS and Atom readers don’t typically authenticate the publisher of each feed every time they download, they might blindly download feeds sent by an impersonating or infected Web publisher, Sima said.
Because RSS and Atom feeds are normally stored by blog readers as HTML files on users’ local hard drives, they also bypass Web browser security settings that might be set to prevent JavaScript from unknown outside Web sites from executing. Turning off JavaScript altogether to prevent such attacks is no solution, Sima said. Too many popular Web sites and applications today rely on JavaScript invisibly executing in the background to operate, making them “painful to use.” The best way to guard against these sorts of attacks, he said, would be for blog-reading software and services to re-encode all JavaScript it receives to render it harmless.
“It could be as simple as adding a few lines of extra code,” he said. Creating this filter would not cause feeds to arrive much slower, either, he said. But as far as he knows, no blog-reading software or service re-encodes the JavaScript as he suggests to de-fang it.
In the absence of blog readers filtering their feeds, Sima recommends that CIOs and chief information security officers start treating individual PCs as potential attack points.
