Windows Defender detection can be bypassed by tricking the antivirus application into scanning a different file or nothing at all, CyberArk Labs reveals.
The technique, which affects the scanning process over SMB shares, allows any malware to bypass Windows Defender and possibly other antivirus applications, researchers from CyberArk say.
Antivirus applications typically catch the execution of an executable file by a kernel callback and then scan the file, usually by requesting the user-mode agent to do so. The operation is different for executables already on the disk compared to those from a SMB share, the researchers explain.
If the executable file is already located on the hard drive, the antivirus won’t scan the process creation, because it scanned the file creation. However, the antivirus would scan the process creation when the executable is run directly from a SMB share, the security researchers explain.
One of the attack vectors involves tricking the antivirus into scanning a different file than the one actually executing. To ensure that one file is served to the Windows PE Loader and another to Windows Defender, a custom implemented SMB server is used.
Thus, when the process creation is made by Windows PE Loader and a request is made to the SMB server for the executable file, a malicious file is delivered. However, when Windows Defender requests the executed file, a benign file is served to ensure that the antivirus doesn’t stop the execution.
Thus, to abuse Windows Defender, an attacker would simply need to implement the SMB protocol and create a “pseudo-server” capable of differentiating between normal requests and those coming from Windows Defender.
One example, the researchers say, would be to decline the oplock request and return STATUS_OPLOCK_NOT_GRANTED, which would result in the scan failing and the malicious file being executed without interruptions. By blocking all handle creation requests with the impersonation level SEC_IDENTIFY one can also block the antivirus from scanning the file.
The attack is possible because the SMB protocol offers transparent integration into Windows, meaning that “accessing a remote file is performed like accessing a local file.” Thus, an attacker would need to create a handle to the file and then perform any operations using specific functions. However, replacing the file is possible for each operation, the researchers say.
CyberArk has contacted Microsoft to report the attack, but the company apparently doesn’t view the issue as a security vulnerability. In fact, the tech giant considers that the various special conditions required to trigger the bypass can be seen as a feature.
“Based on your report, successful attack requires a user to run/trust content from an untrusted SMB share backed by a custom server that can change its behavior depending on the access pattern. This doesn’t seem to be a security issue but a feature,” Microsoft reportedly told CyberArk.