Researchers Find Multiple Critical Flaws In Confide Secure Messaging App

Confide, a messaging app that has seen big growth recently following reports of its popularity for sharing confidential information among Trump aides, may in fact have not been as secure as many of its users may have imagined.

Recent tests conducted by security vendor IOActive turned up multiple critical vulnerabilities in the application, including some which allowed encrypted messages to be intercepted and altered.

All of the vulnerabilities have been fixed after IOActive reported them to Confide. But the discovery of the bugs in an app touted as offering secure, confidential messaging is another reminder of why there is no such thing as bulletproof security on the Internet.

In an emailed statement to Dark Reading, Confide president Jon Brod acknowledged the issues and said they had been addressed. “As a confidential messenger, privacy and security is at the heart of everything we do,” Brod said.

“Our security team is continuously monitoring our systems to protect our users' integrity, and we were able to detect anomalous behavior and remediate many of the issues during IOActive's testing in real time.”

Brod said there is no indication that the flaws, which IOActive had discovered, were exploited.

Confide is best known for being an app that allows users to send encrypted messages that self-destruct after they are read once. The app touts military grade end-to-end encryption and security features like one that prevents screenshots of messages from being taken.

In an interview with CNBC in February, Brod claimed the app’s user base had grown sevenfold in one week after reports surfaced about political operatives in Washington D.C. using it to share information confidentially. One of the reports was from the Washington Post, which claimed that White House staffers, fearful of being identified as speaking with the media, had resorted to using the app to communicate secretly.

IOActive security researchers found the vulnerabilities in Confide version 1.4.2 for Windows and OS X, and version 4.0.4 for Android by reverse engineering the software and observing its behavior and interactions with the public API.

The exercise revealed problems with Confide’s notification, messaging, and account management systems and also with the application’s website as well.

The vulnerabilities gave attackers an opportunity to hijack account sessions or to guess passwords so they could impersonate other users. The flaws also let attackers acquire the contact details of specific Confide users, or all of them, or to launch man-in-the-middle attacks for intercepting and decrypting messages or for altering them without decrypting.

In a security advisory, IOActive said at the time of its testing, Confide’s notification system did not require a valid SSL server certificate, thereby giving attackers opportunity to grab session information via a man-in-the-middle attack.

Another vulnerability, exploited by an attacker using the Confide API directly, would have caused unencrypted messages to be transmitted with message recipients having no indication they had received an unencrypted message, says Ryan O'Horo, a security researcher at IOActive.

In addition, the Confide API would return metadata for certain functions such as "lookup" in order to populate fields in the client user interface, O’Horo says.

“The ‘block’ function, for example, would allow you to block a user based on an incrementing database record number and a sibling function ‘blocked’ would return metadata for users previously specified with ‘block,’” he says.

The Confide application, as tested, gave attackers a way to pull Confide user accounts, including real names, e-mail addresses, and phone numbers, according to IOActive. It allowed user account passwords to be guessed via brute-force methods and did not require users to choose complex passwords.

In all, the IOActive researchers were able to pull some 7,000 records for users who had registered for the app during a two-day period in February. The records suggested that between 800,000 and 1 million user records were stored in the app’s database.

“Although the app had security vulnerabilities when we tested it, this was a good example of a research team and vendor working collaboratively through a successful responsible vulnerability disclosure process,” O’Horo says.

Related stories: