Risk-Based Authentication Practices
A follow-up on Dan’s 2008 Password Manifesto
On the NIST Publication on Digital Identity Guidelines
Another one of those Dan’s New Leaf Posts, meant to inspire thought about IT Governance . . . .
In June 2017, NIST released a special publication on digital identity, NIST SP 800-63, that is starting to get the attention of our larger Clients. We believe this is due in-part because some in management want to use the guidance to justify weaker authentication controls. I posit this because one of the questions I received about this actually pointed to the recommendation to use 8-character passwords on low-risk systems.
Now, I still maintain it’s easier to type a 20-character sentence than an 8-character strong password, but putting that aside, I guess I can see weaker requirements on low-risk systems as okay. But why confuse the heck out of users?? And who has low-risk systems?
But again, let’s put that to the side and talk about real-world uses of a risk-based approach to authentication.
While I admit I didn’t read the entire thing yet (it’s 100+ pages), I do like the notion of taking a “risk-based approach” to authentication. The NIST publication appears to define 3 specific levels of ‘authentication assurance’. Similar to how the CAT now requires us to perform a criticality analysis so we know where to focus our monitoring controls, this guidance defines acceptable authentication requirements for 3 levels of asset sensitivities. They are called Authenticator Assurance Levels, (AAL), where
- AAL1 is for low-risk
- AAL2 is for moderate-risk
- AAL3 is for high-risk
It defines permitted authentication types (AAL1 is anything from just a memorized password to multi-factor cryptographic device whereas AAL3 only allows multi-factor methods), how often re-authentication should be required, and other standards.
In terms of how regulators are implementing it, the auditors at infotex have NOT seen anything referencing these new guidelines in our audits, to-date. Still, since this guidance is relatively new, something will most likely be coming especially for banks over 5 billion.
Be sure to note that the NIST guidelines explicitly state they don’t take into account physical access controls. Additionally, when adding additional authentication factors like biometrics, the need for excessively long and overly complex passwords does reduce a bit (but not entirely). Similar to how we’ve seen guidance change on frequency of password expiration/changes . . . the reason a risk-based approach is great is because it will allow us to base strategy on how people actually use these authentication systems. By requiring unique, excessively long, and complex passwords that are changed too frequently, we may inadvertently be encouraging bad practices like using easily identifiable words and information since users have no other way to keep all these credentials memorized.
For resources, I direct you to my original article on password aging from long before Microsoft finally agreed it was an outdated control: https://my.infotex.com/manifesto/
And, if you haven’t thought to use it, consider sharing this oft reference XKCD comic with your management team, it helps explain how password complexity is sometimes a misnomer in terms of actual strength of passwords: https://xkcd.com/936/
Original article by Dan Hadaway CRISC CISA CISM. Founder and Managing Partner, infotex
Dans New Leaf is a fun blog to inspire thought in the area of IT Governance.