Operations

6/22/2018
01:20 PM
Connect Directly
Twitter
LinkedIn
Google+
RSS
E-Mail
50%
50%

Cracking Cortana: The Dangers of Flawed Voice Assistants

Researchers at Black Hat USA will show how vulnerabilities in Microsoft's Cortana highlight the need to balance security with convenience.

Security vs. convenience is a delicate balance to strike with new technology designed to make our lives easier. Vulnerabilities in voice assistants like Microsoft's Cortana and Amazon's Alexa are perfect examples of how the rush to simplify can cause complicated problems down the road.

Consider Cortana, which is enabled by default in Windows 10. Cortana was built to facilitate vocal interaction with laptops, desktops, smartphones, and IoT devices running Microsoft's newest OS, and it's becoming more common in the enterprise as organizations deploy Windows 10 across their environments. The rollout is leaving holes in enterprise defense because Cortana, like other voice assistants, prioritizes users' comfort over security.

Amichai Shulman, cofounder and CTO of Imperva, has spent the past year exploring the technology, infrastructure, and protocols of voice assistants with his colleague, security researcher Tal Be'ery. He says Cortana serves as a weak interface over the otherwise secure Windows operating system and has discovered multiple vulnerabilities to support this.

"Cortana takes a stable environment, reasonably secured, and adds this new type of interface on top of it," Shulman explains. He and Be'ery will present their findings on Cortana flaws in August at Black Hat USA in a session entitled "Open Sesame: Picking Locks with Cortana."

Want Data? All You Have to Do Is Ask

It didn't take long for the researchers to learn the default setting in Windows 10 lets anyone communicate with Cortana – even when the machine is locked. "We started looking for methods to bypass the lock mechanism using voice," says Shulman.

They first found the Voice of Esau (VoE) exploit, which let attackers take over a locked Windows 10 device by combining voice commands with network manipulation to put a malicious payload on a target machine.

VoE was patched by Microsoft in August 2017; still, Shulman believed there were more vulnerabilities to be uncovered. He continued his research, testing concepts like vocal malware along with students at Technion, the Israel Institute of Technology. There are local and Internet-based extensions for Cortana; a whole set of protocols and APIs that can be attacked, he says.

One of the flaws they discovered, dubbed Open Sesame, will be the core focus of their Black Hat session. CVE-2018-8140 was recently addressed in Microsoft's June Patch Tuesday update.

Open Sesame is more powerful than VoE, he says, and allows attackers to take over a locked Windows 10 machine and execute arbitrary code. Exploiting this vulnerability would let anyone view sensitive files, browse online, download and launch arbitrary executables from the Internet, and in some cases gain elevated privileges. A threat actor wouldn't need external code to pull this off, so antivirus and IPS tools wouldn't pick up on the attack.

"There are a number of scenarios where close proximity attacks are being used – computers in hotel rooms, airport security attacks … just give your locked computer to someone to inspect and in a matter of seconds that computer can be compromised," Shulman explains.

A Conversation on Convenience

The reason behind these vulnerabilities is twofold, Shulman says. For starters, Microsoft tried to pitch a new interface, built on accessibility and comfort, onto a secured multi-purpose computing environment. "When you do that, there are glitches," he points out.

It is beneficial to have some commands available on a locked machine. Asking about the weather, for example, is benign. However, it's tricky to enable certain functionalities while denying operations that could potentially compromise the machine.

The second problem: Microsoft's assistant is more than a voice interface. Shulman calls it "an intent resolution engine." Cortana takes natural language input and conveys it into an action performed by the machine. However, it accepts both written and spoken commands. Once the system is invoked with "Hey Cortana," an attacker can enter keyboard input and execute code on a locked device.

As part of their discussion, Shulman and Be'ery will address defense mechanisms for protecting against these types of attacks. One lesson, at least for enterprise environments, is to make sure the enhanced interface is off by default. Vendors don't like to launch new services and not have them active by default, Shulman says, but in the workplace "it's the prudent thing to do."

The team will demonstrate the Open Sesame vulnerability at Black Hat There are three more Cortana vulnerabilities Shulman hopes to discuss in their presentation but they are waiting on confirmation before they can publicly share more details.

Related Content:

Why Cybercriminals Attack: A DARK READING VIRTUAL EVENT Wednesday, June 27. Industry experts will offer a range of information and insight on who the bad guys are – and why they might be targeting your enterprise. Go here for more information on this free event.

 

Kelly Sheridan is the Staff Editor at Dark Reading, where she focuses on cybersecurity news and analysis. She is a business technology journalist who previously reported for InformationWeek, where she covered Microsoft, and Insurance & Technology, where she covered financial ... View Full Bio

Comment  | 
Print  | 
More Insights
Comments
Newest First  |  Oldest First  |  Threaded View
White House Cybersecurity Strategy at a Crossroads
Kelly Jackson Higgins, Executive Editor at Dark Reading,  7/17/2018
The Fundamental Flaw in Security Awareness Programs
Ira Winkler, CISSP, President, Secure Mentem,  7/19/2018
Number of Retailers Impacted by Breaches Doubles
Ericka Chickowski, Contributing Writer, Dark Reading,  7/19/2018
Register for Dark Reading Newsletters
White Papers
Video
Cartoon Contest
Write a Caption, Win a Starbucks Card! Click Here
Latest Comment: This comment is waiting for review by our moderators.
Current Issue
Flash Poll
The State of IT and Cybersecurity
The State of IT and Cybersecurity
IT and security are often viewed as different disciplines - and different departments. Find out what our survey data revealed, read the report today!
Twitter Feed
Dark Reading - Bug Report
Bug Report
Enterprise Vulnerabilities
From DHS/US-CERT's National Vulnerability Database
CVE-2018-14492
PUBLISHED: 2018-07-21
Tenda AC7 through V15.03.06.44_CN, AC9 through V15.03.05.19(6318)_CN, and AC10 through V15.03.06.23_CN devices have a Stack-based Buffer Overflow via a long limitSpeed or limitSpeedup parameter to an unspecified /goform URI.
CVE-2018-3770
PUBLISHED: 2018-07-20
A path traversal exists in markdown-pdf version <9.0.0 that allows a user to insert a malicious html code that can result in reading the local files.
CVE-2018-3771
PUBLISHED: 2018-07-20
An XSS in statics-server <= 0.0.9 can be used via injected iframe in the filename when statics-server displays directory index in the browser.
CVE-2018-5065
PUBLISHED: 2018-07-20
Adobe Acrobat and Reader 2018.011.20040 and earlier, 2017.011.30080 and earlier, and 2015.006.30418 and earlier versions have a Use-after-free vulnerability. Successful exploitation could lead to arbitrary code execution in the context of the current user.
CVE-2018-5066
PUBLISHED: 2018-07-20
Adobe Acrobat and Reader 2018.011.20040 and earlier, 2017.011.30080 and earlier, and 2015.006.30418 and earlier versions have an Out-of-bounds read vulnerability. Successful exploitation could lead to information disclosure.