Microsoft tries to protect user account credentials from theft in Windows 10 Enterprise, and security products detect attempts to pilfer user passwords. But all those efforts can be undone by Safe Mode, according to security researchers.
The Safe Mode is an OS diagnostic mode of operation that has existed since Windows 95. It can be activated at boot time and only loads the minimal set of services and drivers that Windows requires to run.
This means that most third-party software, including security products, don't start in Safe Mode, negating the protection they otherwise offer. In addition, there are also Windows optional features like the Virtual Secure Module (VSM), which don't run in this mode.
VSM is a virtual machine container present in Windows 10 Enterprise that can be used to isolate critical services from the rest of the system, including the Local Security Authority Subsystem Service (LSASS). The LSASS handles user authentication. If VSM is active, not even administrative users can access the passwords or password hashes of other system users.
On Windows networks, attackers don't necessarily need plaintext passwords to access certain services. In many cases the authentication process relies on the password's cryptographic hash, so there are tools to extract such hashes from compromised Windows machines and use them to access other services.
This lateral movement technique is known as pass-the-hash and is one of attacks that Virtual Secure Module (VSM) was intended to protect against.
However, security researchers from CyberArk Software realized that since VSM and other security products that could block password extraction tools don't start in Safe Mode, attackers could use it to bypass defenses.
Meanwhile, there are ways to remotely force computers into Safe Mode without raising suspicions from users, CyberArk researcher Doron Naim said in a blog post.
To pull off such an attack, a hacker would first need to gain administrative access on the victim's computer, which is not that unusual in real-world security breaches.
Attackers use various techniques to infect computers with malware and then escalate their privileges by exploiting unpatched privilege escalation flaws or by using social engineering to trick users.
Once an attacker has admin privileges on a computer he can modify the OS's boot configuration to force it to automatically enter Safe Mode the next time it's started. He can then configure a rogue service or COM object to start in this mode, steal the password and then reboot the computer.
Windows normally displays indicators that the OS is in Safe Mode ,which could alert users, but there are ways around that, Naim said.
First, to force a reboot, the attacker could display a prompt similar to the one shown by Windows when a computer needs to be restarted to install pending updates. Then once in Safe Mode, the malicious COM object could change the desktop background and other elements to make it seem that the OS is still in normal mode, the researcher said.
If attackers want to capture a user's credentials, they need to let the user log in, but if their goal is only to execute a pass-the-hash attack, they can simply force a back-to-back restart which would be indistinguishable to the user, Naim said.
CyberArk reported the issue, but claims that Microsoft doesn't view it as a security vulnerability because attackers need to compromise the computer and gain administrative privileges in the first place.
While a patch might not be forthcoming, there are some mitigation steps that companies could take to protect themselves against such attacks, Naim said. These include removing local administrator privileges from standard users, rotating privileged account credentials to invalidate existing password hashes frequently, using security tools that function properly even in Safe Mode and adding mechanisms to be alerted when a machine boots in Safe Mode.