After previously discovering other vulnerabilities in the Kaspersky antivirus program, Google’s Project Zero security researcher, Tavis Ormandy, has found another puzzling one. The issues lies with Kaspersky’s interception of HTTPS traffic with its own certificate in order to scan for web threats.
Antivirus TLS Interception
The Kaspersky antivirus, just like a few other other antivirus tools, offers users the option (sometimes enabled by default) to allow it to scan within TLS/HTTPS connections, too. Before websites started moving to HTTPS encryption, antivirus or other web analysis tools could just look at the traffic as it was coming into the browser. However, that’s not possible anymore with websites that have encrypted their traffic.
To continue to analyze that traffic, these tools would need to install their own certificates on the users’ computers, and then decrypt that traffic. This is similar to how man-in-the-middle attacks happen. However, the difference is that presumably the users are aware this is happening when they enable the web scanning option in their antivirus programs and that they trust the antivirus companies not to do nefarious things with their web activities.
In practice, it’s likely that most users aren’t aware the antivirus software can see their traffic, though, because not everyone is aware of all the intricacies of antivirus and security technologies. Therefore, this “solution” to encrypted web threats may unnecessarily put most users at risk when they don’t know what they’re doing. Some security experts such as Thomas Ptacek believe no antivirus program should be allowed to do TLS interception of all web traffic.
Kaspersky's New Vulnerability
According to Ormandy, the Kaspersky antivirus would install its own root certificate on the user’s computer, and doing it in a way that’s not well protected, either. It would then replace all the visited websites’ certificates with its own generated leaf certificates. So far, this is expected behavior for web scanning tools (although users should still be wary of which tool they allow to do this on their computers).
The problem that Ormandy discovered is that Kaspersky was re-using 32-bit keys for its leaf certificates. This would make it easy for an outside attacker to brute-force a collision and intercept the traffic of multiple sites when Kaspersky users would access them.
The bug would either not allow users to connect to the websites, or the websites would be downgraded to unencrypted HTTP connections, thus allowing attackers to potentially intercept the connections.
According to the Google engineer, an attack would go like this:
- Mallory wants to intercept mail.google.com traffic, for which the 32bit key is 0xdeadbeef.
- Mallory sends you the real leaf certificate for mail.google.com, which Kasperksy validates and then generates its own certificate and key for.
- On the next connection, Mallory sends you a colliding valid certificate with key 0xdeadbeef, for any commonName (lets say attacker.com)
- Now mallory redirects DNS for mail.google.com to attacker.com, Kaspersky starts using their cached certificate and the attacker has complete control of mail.google.com.
Ormandy thought it was “incredible” that the Kaspersky team wouldn’t notice that sometimes they would get certificate errors, even if by accident. However, he doesn’t venture to say that Kaspersky may have implemented this vulnerability on purpose.
Ormandy noted that because Google is using its new open source encrypted transport protocol called QUIC for its own services, when accessed from Chrome, Kaspersky is in fact not able to decrypt the Google services connections in Chrome, but it can do it in Firefox or other browsers.
Google's researcher told Kaspersky about the vulnerability on November 1, and the typical 90-day disclosure policy applied. The bug was fixed by the time the 90 days passed. Therefore this particular vulnerability may not put users at risk anymore, but all the issues with antivirus TLS interception remain.
Does Antivirus Software Make You Safer?
The primary reason for using an antivirus tool is to protect yourself against malware that takes advantage of existing and well known vulnerabilities. Some, or perhaps many users don’t update their operating systems or applications on time, which leaves them vulnerable to malware. In such cases, antivirus software could serve a good role of keeping those users safe.
However, for users who always update their operating systems and applications on time, an anti-virus software is much less necessary, because the vulnerabilities that would normally be exploited by malware have been closed.
There are “zero-day” vulnerabilities as well, of course, and by the very definition there's no patch. However, anti-virus software can’t protect you from zero-day vulnerabilities either. Some of them, including the Kaspersky Antivirus, do offer some limited anti-exploit protections, but you may have to balance that with the fact that the antivirus itself may make you less secure.
There likely isn’t a definite answer to whether antivirus software makes you safer. For people who can’t or won’t update their systems, chances are that antivirus software does make them safer. For others, who are always up to date, it may be worth investing more in specialized anti-exploit tools, or even better, sandboxing and other virtualization technologies that wouldn’t allow malware to hurt the main operating system.