Researchers from security firm FireEye claim that attackers are actively using a remote code execution exploit that works against the latest versions of Adobe Reader 9, 10 and 11.
"Today, we identified that a PDF zero-day [vulnerability] is being exploited in the wild, and we observed successful exploitation on the latest Adobe PDF Reader 9.5.3, 10.1.5, and 11.0.1," the FireEye researchers said late Tuesday in a blog post.
The exploit drops and loads two DLL files on the system. One file displays a bogus error message and opens a PDF document that's used as a decoy, the FireEye researchers said.
Remote code execution exploits regularly cause the targeted programs to crash. In this context, the fake error message and second document are most likely used to trick users into believing that the crash was the result of a simple malfunction and the program recovered successfully.
Meanwhile, the second DLL installs a malicious component that calls back to a remote domain, the FireEye researchers said.
It's not clear how the PDF exploit is being delivered in the first place -- via email or over the Web -- or who were the targets of the attacks using it. FireEye did not immediately respond to a request for additional information sent Wednesday.
"We have already submitted the sample to the Adobe security team," the FireEye researchers said in the blog post. "Before we get confirmation from Adobe and a mitigation plan is available, we suggest that you not open any unknown PDF files."
The Adobe Product Security Incident Response Team (PSIRT) confirmed Tuesday in a blog post that it is investigating a report of a vulnerability in Adobe Reader and Acrobat XI (11.0.1) and earlier versions being exploiting in the wild. The risk to customers is being assessed, the team said.
In response to a request for a status update sent Wednesday, Heather Edell, Adobe's senior manager of corporate communications, said that the company is still investigating.
Sandboxing is an anti-exploitation technique that isolates a program's sensitive operations in a strictly controlled environment in order to prevent attackers from writing and executing malicious code on the underlying system even after exploiting a traditional remote code execution vulnerability in the program's code.
A successful exploit against a sandboxed program would have to leverage multiple vulnerabilities, including one that allows the exploit to escape from the sandbox. Such sandbox bypass vulnerabilities are rare, because the code that implements the actual sandbox is usually carefully reviewed and is fairly small in length compared to the program's overall codebase that could contain vulnerabilities.