This blog post is republished with the permission of The Cipher Brief. See the original post here.
Last fall brought news that the victims of the OPM fingerprint breach expanded to over five million prints. It’s for this reason that the safety of biometric data should be questioned and discounted as a viable means for authentication. Multiple techniques are available for using this type of information to create fake fingerprints to bypass biometric scanners, plant false fingerprints, or even falsify applications that need fingerprint data using traditional ink techniques. While vendors gather around biometrics as a holy grail for authentication, it is breaches like this that put the entire concept of biometrics-based security in jeopardy for the masses. Therefore, what is needed to solve this problem is a clear definition of when biometrics should be used for authorization, authentication, and to support two-factor authentication. While these may sound like similar terms, in reality, biometrics should only be used for authorization and never authentication alone.
Authorization, in the simplest terms, is the permission to perform a task. It is the ability to proceed without verifying who you are, or who you say you are. The most common form of biometric authorization used today is Apple Pay. When placing your finger on the touch identification sensor, you are authorizing payment. It is just a permission. Authentication, however, is the verification of you as a person, and who you say you are. It does not authorize you to perform any tasks; it just proves your identity. Authentication is primarily performed today by usernames and passwords, two-factor authentication, smart cards, and other techniques like one-time-passwords. They generally tie secret knowledge to a second physical media or to the creation of a unique code that only you have knowledge of. The various components of an authentication system are designed to prove your identity, but they do not authorize you as a person to anything.
So here is where the problem lies. Newer biometric technologies are blurring the line between authorization and authentication. They are taking a very sophisticated approach based on technology to identify an individual (authentication) and are now merging it with the permissions to perform a task (authorization). When biometric data is compromised, both suffer and can be used to infiltrate either the user’s identity or the ability to perform tasks. While OPM is the first major breach of this type, it does have far-reaching ramifications. The biometric data stolen can be used to craft authorization and authentication attacks against anyone breached, including some of the highest valued personas and assets in the world. Either type of attack would easily and completely bypass traditional username and password paradigms if biometrics alone where used for both, as some vendors are proposing.
This is why there needs to be a separation between authentication and authorization, and why the definitions between the two need to be clearly understood if biometrics is going to succeed in authorization. Biometric data is like any other form of data. It gets stored electronically. Granted, it is encrypted, and it uses various forms of key mechanisms to ensure that one data source alone cannot reveal the contents, but that is not the point. Vendors claim the data is not hack-able, but history has shown us, from the Enigma machine to RSA Keys, that every type of encryption can be compromised and probably will be. It is just a matter of time, persistence, and a fast computer. While there are systems today that are impossible to hack, that claim only has a finite lifetime. Biometric data is stuck with you for a lifetime. You cannot change your fingerprints or the blood vessels in your eyes. So, by the time you’re old, there is a good chance that system securing your biometrics might not be as secure as it is today. Considering we have computers from the 1950’s running air traffic flight control, power plants, and other infrastructure, it is reasonable to assume some of today’s systems and databases may also be around in 60 years.
There is one final discussion point that’s worth noting: The debate regarding biometrics is not new. The ability to steal someone’s likeness is not limited to science fiction and has been documented over and over again via the Internet and TV shows like MythBusters. What is new is the extent hardware and software vendors are pushing biometrics as the next generation of security solutions. In the last 10 years, we have seen fingerprint readers on laptops (most of which have been cracked) and facial recognition software which uses local cameras to verify user identities as part of a login prompt; all of which have come and gone. This next generation push of biometrics uses multiple techniques for facial recognition (visual, infrared, etc.) and ultimately stores the data electronically, just like every other previous technology. They are computers with storage devices—encrypted or not—and biometric information still needs to be referenced somehow to complete the authentication process. Thus, somehow, sometime, they will be compromised.
The OPM breach has shown it is possible to steal biometric data. In 2016, the hype around this technology will implode when that data is used against our own assets and information. Consumers and businesses adopting biometrics need to be aware of the separation of tasks and protect accordingly. The implosion of this technology is not far off if we continue to blur the lines of authorization and authentication. As long as we can keep the two separate, the risk can be managed and the future predictions of identity chaos can be avoided.