Security researchers have highlighted in recent months how the web proxy configuration in browsers and operating systems can be abused to steal sensitive user data. It seems that attackers are catching on.
A new attack spotted and analyzed by malware researchers from Microsoft uses Word documents with malicious code that doesn't install traditional malware, but instead configures browsers to use a web proxy controlled by attackers.
In addition to deploying rogue proxy settings, the attack also installs a self-signed root certificate on the system so that attackers can snoop on encrypted HTTPS traffic as it passes through their proxy servers.
One of the PowerShell scripts deploys a self-signed root certificate that will later be used to monitor HTTPS traffic. Another script adds the same certificate to the Mozilla Firefox browser, which uses a separate certificate store than the one in Windows.
The third script installs a client that allows the computer to connect to the Tor anonymity network. That's because the attackers use a Tor .onion website to serve the proxy configuration file.
The system's proxy auto-config setting is then modified in the registry to point to the .onion address. This allows attackers to easily change the proxy server in the future if it's taken offline by researchers.
"At this point, the system is fully infected and the web traffic, including HTTPS, can be seen by the proxy server it assigned," the Microsoft researchers said in a blog post. "This enables attackers to remotely redirect, modify and monitor traffic. Sensitive information or web credentials could be stolen remotely, without user awareness."
Researchers from the SANS Internet Storm Center recently reported a similar attack from Brazil, where hackers installed rogue proxies on computers in order to hijack traffic to an online banking website. A rogue root CA certificate was deployed in that case as well in order to bypass HTTPS encryption.
At the DEF CON and Black Hat security conferences earlier this month, several researchers showed how man-in-the-middle attackers can abuse the Web Proxy Auto-Discovery (WPAD) protocol to remotely hijack people's online accounts and steal their sensitive information, even when those users access websites over encrypted HTTPS or VPN connections.