Companion mobile app exposed Hyundai cars to potential hijacking

The mobile application that accompanies many Hyundai cars exposed sensitive information that could have allowed attackers to remotely locate, unlock, and start vehicles.

The vulnerability was patched in the latest version of the mobile app released in March but was publicly disclosed on Tuesday. It is the latest in a string of flaws found over the past few years in the “smart” features added by vehicle manufacturers to their cars.

The Hyundai issue was discovered by independent researchers William Hatzer and Arjun Kumar when analyzing the MyHyundai with Blue Link mobile app.

NEWSLETTERS: Get the latest tech news sent directly to your in-box

Blue Link is a subscription-based technology that’s available for many Hyundai car models released after 2012. It allows car owners to remotely locate their vehicles in case of theft, to remotely unlock them if they lose or misplace their keys, and even to remotely start or stop their engine when they’re parked and locked.

The researchers found that starting with version 3.9.4 of the mobile app, released in early December, Hyundai added a feature to upload a log file to a remote server. The connection to the server was not encrypted with HTTPS, but Hyundai attempted to secure the log data by encrypting it with a static key shared by all users.

The problem is that this key is hard-coded inside the application, so anyone can download the app, find the key, and extract it. This allows an attacker who is in a position to intercept traffic from a mobile phone with the application installed to intercept and decrypt the log data.

The application log contains sensitive information like the car owner’s username, password, and PIN, as well as GPS data that could disclose the car’s location history.

Man-in-the-middle attacks can be executed on insecure wireless networks, through compromised routers, or even higher up in the network chain, if the attacker gains visibility at the ISP level.

Hatzer and Kumar worked with security firm Rapid7 to coordinate the vulnerability disclosure with Hyundai, the U.S. Department of Homeland Security’s Industrial Control Systems Cyber Emergency Response Team (ICS-CERT), and the CERT Coordination Center at Carnegie Mellon University.

ICS-CERT published an advisory for the issue Tuesday, rating the failure to secure communications against MitM attacks as medium severity and the use of a hard-coded encryption key as high severity.

Hyundai Motor America fixed the flaw in version 3.9.6 of its Blue Link mobile app for Android and iOS. The company told Rapid7 that following its investigation it is not aware of any customers affected by this issue.

Hyundai is a member of the Automotive Information Sharing and Analysis Center (Auto-ISAC), an industry group for sharing information about cybersecurity threats and best practices.

There has been some progress in automotive cybersecurity over the past few years, with several manufacturers launching bug bounty programs and showing their willingness to collaborate with security researchers. However, it will take some time until those efforts start having a significant impact on the quality of car-related software.

To some extent, this is understandable, because car companies are relatively new to software development. Large software vendors like Microsoft, Adobe, and Oracle are still finding and fixing dozens of security flaws in their products every month, so it’s not surprising that car software or accompanying mobile apps also have vulnerabilities.

There is some urgency, though, because compared to computers, cars are a bigger threat to human safety. And unfortunately, the flaws being discovered in car software are often basic security oversights that could have easily been avoided by following well-known secure development principles.

In another recent incident, researchers from Israeli firm Argus Cyber Security were able to turn off the engine of a moving car when in Bluetooth range due to a vulnerability in a car monitoring device from Bosch called the Drivelog Connector. This dongle connects to a car’s on-board diagnostics (OBD-II) port and monitors its “health” to alert drivers when the car needs servicing. It comes with an accompanying mobile app.

The Argus researchers found an issue in the Bluetooth-based authentication system between the app and the dongle that allowed them to guess the authorization PIN by using brute-force methods after attempting to connect to the dongle once. With the PIN, the researchers found a way to remotely send malicious commands through the dongle to the controller area network (CAN Bus) that’s used by the car’s electronic control units (ECUs) to communicate.

The CAN Bus doesn’t have any encryption or authentication because ECUs need to communicate with each other and react to sensor readings as fast as possible in order to perform critical functions. Because of that, gaining unrestricted access to the CAN Bus is the holy grail for any car hacker.

Unfortunately, previous car hacks have shown that car manufacturers don’t properly isolate the CAN Bus from peripheral and remotely accessible systems. In 2015, researchers Charlie Miller and Chris Valasek found a way to hack into the infotainment system of the Jeep Cherokee and other Fiat Chrysler vehicles over a mobile data connection and then jump to the CAN Bus to take over the brake, steering, and other critical systems.

Making cars “smarter” by adding internet connectivity and remotely accessible features can certainly enhance a car owner’s experience, but it also increases the cars’ attack surface. It’s possible that many of these functions will be properly secured in time if car makers continue to take cybersecurity seriously, but for now, it seems more likely that if a car has remote control features, there’s an exploitable bug somewhere in the implementation.

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.