Messaging programs are a closely watched application category, with experts scrutinizing how communications are protected from government surveillance dragnets and hackers. The primary defense invariably involves encryption, but just saying an application uses encryption by no means ensures it’s secure.
One of the latest programs to come under fire is Telegram, which is backed by Pavel Durov[cq], who also founded the popular Russian social networking site Vkontakte. Telegram is a free desktop and mobile application launched in 2013 that promotes itself as “taking back our right to privacy.”
Telegram is well intended but has several weak spots, said Alex Rad[cq], who has a background in application security testing and reverse engineering. He and researcher Juliano Rizzo, who discovered two major attacks against SSL (Secure Sockets Layer), have been analyzing Telegram intermittently since last year as a side project to help improve its security.
They went public on Sunday with a blog post pointing out problems with Telegram, which may cause concern for those who are particularly worried about how such messaging systems could be compromised. Rad said in a phone interview that his correspondence with Telegram has been cordial but a bit tense.
“What bothered me about Telegram was the way they market themselves versus the reality of how people use their application,” said Rad, who lives in Stockholm.
For example, Telegram doesn’t implement end-to-end encryption by default, a technique that ensures a message is encrypted on a device and is only decrypted by a recipient. That kind of encryption is regarded as the safest way to send information.
To send a fully encrypted message, Telegram users must initiate a “secret chat.” But Rad said there are potential problems with how a secret chat is set up that could make it vulnerable to a man-in-the-middle (MITM) attack.
Before a secret chat begins, two Telegram users see an image that verifies their connection hasn’t been tampered with. Rad describes in the blog post how an attacker could replace that image with one of their own, potentially giving assurance to users that their chat is secure when it is not.
Determining whether the MITM attack would even be feasible leads to an academic argument about computing power. Telegram has dismissed the attack in a blog post as too expensive to pull off. It also requires that the attacker already has access to Telegram’s servers, an assumption that Rad concedes makes a MITM attack on two users less likely given the vast hacking opportunities that such a position would afford anyway. But he also said his theoretical attack could be made impossible by using a stronger encryption algorithm, a trivial upgrade for Telegram.
Telegram’s Markus Ra[cq] said via email that while his company contests the feasibility of Rad’s attack, “Telegram’s secret chats are evolving constantly, and we’ll make sure they stay secure even as potential attackers gain processing power over time.”
Rad said there’s also a potential problem with Telegram’s user authentication. When a person registers a new device, Telegram sends a one-time passcode via SMS. There are a variety of attacks and methods to intercept SMS communications, Rad said, making it generally not the best way to authenticate a user.
Telegram’s Ra conceded issues around SMS and wrote that Telegram has been working on an additional cloud-based password mechanism for the last two months. “This work is now almost done and our users should be able to enjoy two-factor authentication soon,” he wrote.