Cisco Webex today patched three vulnerabilities in its videoconferencing platform that could allow an attacker to join meetings as a "ghost" without appearing on the participant list, stay in a meeting after being expelled, and gain access to attendees' data in the meeting room lobby.
The vulnerabilities were discovered by IBM Research, where experts decided to analyze the software they were depending on most as employees transitioned to home offices, says Ian Molloy, principal RSM and department head for IBM Research.
"Our CISO recommended looking into best practices around meetings – best ways to protect the company," he says.
Webex, which is IBM's primary tool for remote meetings, became the research subject. The vulnerabilities they discovered exist in the "handshake" process that Webex uses to create connections between meeting attendees, researchers explained in a writeup of their findings. As part of this process, a client system and server exchange "join" messages with data on meeting attendees, the client application, meeting ID, meeting room details, and other information.
An attacker with a meeting URL could manipulate the messages between the Webex client application and Webex server back-end to join, and stay in, a meeting without being seen by other attendees. Researchers say they identified the specific pieces of client data that an attacker would need to manipulate to sneak into a Webex meeting undetected.
When a host starts or unlocks a call, an attacker could use the handshake manipulation to slip in unbeknown to attendees. They would be able to see and hear other participants, view shared screens, and chat without revealing their presence. Attendees may hear an extra beep when the extra person sneaks in; other than that, their presence would likely go unnoticed – unless the ghost started to chat with other participants or send messages to the group.
The Ghost Join flaw (CVE-2020-3419) would let an uninvited guest join a Webex meeting, while a separate vulnerability (CVE-2020-3471) would let them stay after a host kicked them out. The host and other attendees would not see the ghost on a participant list, but they may still listen.
"With increasing back-to-back meetings, this tactic would allow an attacker to listen to more sensitive conversations and could be used in conjunction with social engineering to join locked meetings," researchers wrote, noting once someone becomes a ghost attendee, it's impossible to see them.
The third vulnerability (CVE-2020-3441) could benefit an attacker before the meeting starts. In the meeting lobby, they could access participant data such as name, email address, IP address, how they connected (phone, browser, Webex Room Kit), and other system details. This data was accessible even if a meeting was locked or hadn't yet started, the researchers noted.
"During the communication protocol, we noticed the participants' name, email, and the IP address are all being collected, even in the lobby," says research scientist Jiyong Jang. "So that means you can potentially collect who might be entering the meeting at this time, and then you can see their location from their IP address."
This data could be used for further reconnaissance or in more targeted attacks, he adds. The IP address was concerning for employees in home offices, as it revealed the ISP, geolocation, and consumer-grade home network, which are usually less secure than enterprise networks.
"We're not all behind the corporate firewalls and the corporate environments," Molloy says. "We're relying on the security of consumer-grade home routers that your ISP might've given you, where you don't know if everything is as locked down as it should be otherwise."
The researchers were able to demonstrate the "ghost" issue on macOS, Windows, and the iOS version of Webex Meetings applications and Webex Room Kit appliance. Technically, Jang says, it's "not too difficult" to pull off this attack; the tools needed are commonly used at capture-the-flag competitions or among penetration testers, he notes.
Patches available today should be applied, but the researchers shared some of the workaround they adopted while fixes were in the works. IBM teams began to lock meetings at zero minutes, Molloy says, after which attendees must be manually admitted by the host. He advises watching for external attendees who may seem suspicious or for random phone numbers.
"We learned which precautions to take. We changed our own internal habits and started locking our meetings down as best we could," Molloy says.
Jang suggests locking meetings with a passcode, which attackers would then also need to exploit these vulnerabilities. IBM teams also began scheduling meetings with unique IDs instead of personal meeting rooms, which are "easily repeatable but easily guessed," Molloy adds.