Adobe Sandbox Protection Hacked

Last year Adobe got the message that it was the target du jour for malicious attacks and stepped up its efforts to secure and protect its software. However, a security researcher revealed that one of the security measures developed by Adobe can be circumvented with relative ease.

The sandbox security control creates a sort of software safe zone–a segregated area where scripts and code can execute without putting the core functionality of the application at risk or allowing broader access to the PC itself. If potentially malicious software is able to bypass the sandbox, though, it can operate outside of that safe zone and potentially exploit the system or compromise data.

Anup Ghosh, founder and chief scientist for Invincea e-mailed the following comments to me. “When Adobe announced its plans to roll out its Reader X product, we commended them for trying to solve some of the major exploit problems they faced through the introduction of a sandbox. But we warned them that while application sandboxes are a step in the right direction, they leave residual risk on the table that will be exploited in the months to come.”

Apparently, that time has come. Billy Rios, a security researcher, has found that getting around the Adobe sandbox is somewhat trivial for someone who knows what they’re doing. Rios described the hack in a recent blog post that, while the Adobe sandbox does restrict some code execution, “Unfortunately, these restrictions are not the same as, “cannot communicate with the network in any way” which is what is stated in the documentation. The simplest way to bypass the local-with-filesystem sandbox is to simply use a file:// request to a remote server.”

Ghosh explains, “The exploit doesn’t involve breaking any code or finding any flaws in implementation. Rather, it shows the flaw in the design of the sandbox. As with any application sandbox, the sandbox designers are forced into an untenable position of trying to think of every way an attacker may attempt to exploit privileges and then come up with methods (bandages) to block those paths.”

There is some silver lining to the sandbox bypass, though. Rios notes, “Fortunately, it seems you can only pass IPs and hostnames for system on the local network (RFC 1918 addresses). If an attacker wants to send data to a remote server on the Internet we’ll have to resort to a couple other tricks.”

In fairness to Adobe, as well as to sandbox implementations elsewhere, nobody ever claimed that the sandbox is the perfect security solution, or that it is impenetrable. As Rios pointed out, Adobe did use the unfortunate language of stating that files in the sandbox can not communicate with the network in any way–which is false, but just be aware that any security control is just one method or layer of protection and there are no silver bullets.