Original URL: http://www.theregister.co.uk/2013/07/31/whitehat_security_ad_networks_botnet/

Malicious JavaScript flips ad network into rentable botnet

Enslaved machines helplessly press Apache's buttons

By Jack Clark

Posted in Security, 31st July 2013 21:16 GMT

Black Hat 2013 Security researchers have shown how hackers can use ad networks to create ephemeral, hard-to-trace botnets that can perform distributed-denial-of-service attacks at the click of a button.

In a presentation at the Black Hat conference in Las Vegas on Wednesday, researchers from WhiteHat Security showed off their technique, which uses iframes in web ads to call a JavaScript file that hammers a site with requests.

The exploit "forces JavaScript to use cross-origin requests to force as many requests as possible out of a single browser or a lot of browsers to a single website," WhiteHat Security's threat research center chief Matt Johansen said.

The company embedded JavaScript code in an advert that they ran on an unnamed ad network. This code pointed to an Amazon Web Services server on which they hosted the core JavaScript file, which they could then modify after the ad was deployed.

WhiteHat confirmed that the ad network did evaluate the code, but seeing nothing overtly malicious, permitted it to go ahead.

"We had kind of benign JavaScript here, but if you started using the evil ideas the code might start to look a bit suspicious," Johansen said. "We didn't dip our toe into the [ad] porn networks."

The researchers' code asked the browser to throttle up to its maximum amount of connections (six in Firefox, for example) and access the website via HTTP. They also demonstrated a workaround that can go above the browsers' permitted number of concurrent connections by using an FTP request format, potentially allowing one browser to flood a site with concurrent connections.

This approach let the researchers deploy an ad that could automatically execute when served on a page and force viewers' browsers to hammer a site of WhiteHat's choice with requests.

"What's the benefit of hacking this way – why not do a traditional DDoS attack?" asked WhiteHat's threat research center manager Matt Johansen, who then answered his own question. "There is no trace of these. The JS gets served up, it goes away. It's very, very easy."

The only real way to trace this back to WhiteHat would be to go to the ad network and get the credit card used to buy the malicious adverts, Johansen said. As Reg readers will know, it's not too difficult for hackers to illicitly and anonymously gain access to credit cards.

In a live demonstration, the researchers showed 256 concurrent connections to a single Apache Web Server, with over a million connections tracked in an hour. The total cost of the ads was lower than the cost of the Amazon instance used to serve the illicit JavaScript, and both only cost tens of dollars.

Next, WhiteHat plans to work with partners to deploy a version of the exploit that explicitly targets a site protected by a DDoS-protection service. They also plan to try and use the technique to run distributed MD5 hash cracking via a software tool such as Ravan. Previously, the same researchers have cracked open Google's Chrome OS.

Much to the dismay of this ad-funded publication, the researchers plugged the use of ad blockers as one of the only easy ways to remediate this problem. ®