With more and more consumers having their passwords compromised on a daily basis, a pair of researchers are floating an idea that they contend will help foil digital credential crackers.
They propose salting a web-site’s password database with lots of false passwords called “honeywords.” Passwords in password databases are typically “hashed” or scrambled to protect their secrecy.
“An adversary who steals a file of hashed passwords and inverts the hash function cannot tell if he has found the password or a honeyword,” Ari Juels of RSA Labs and MIT Professor Ronald L. Rivest wrote in paper titled Honeywords: Making Password-cracking Detectable that was released last week.
“The attempted use of a honeyword for login sets off an alarm,” they added.
How it would work
A password database salted with honeywords would be plugged into a server dedicated exclusively to distinguishing between valid passwords and honeywords. When it detects a honeyword being used to log into an account, it will alert a site administrator of the event, who can lock the account down.
Using honeywords won’t prevent hackers from breaching your website and stealing your passwords, but it will alert the operators of the website that a breach may have occurred.
“That’s very valuable,” Ross Barrett, senior manager of security engineering at Rapid7 told PCWorld—especially in light of how long it’s taking to uncover many of these breaches.
“The average time for detection of a compromise is six months,” he said, “and that’s increased from last year.”
However, if hackers knew a site was using honeywords and accounts are automatically locked down when a honeyword is used, the honeywords could actually be used to create a denial-of-service attack on the site.
The scheme also gives attackers another potential target: the honeychecker. If communications between the checker and the web-site server is disrupted, the web site could crash.
Even if the honeychecker is compromised, though, a web-site operator is still better off with honeywords than without them, Barrett contended.
“It was probably a lot harder for those hackers to break into their website than if they hadn’t had a honeychecker,” he said.
Juels and Rivest recommend in their paper that the honeychecker be separated from the computer systems running a web site.
“The two systems may be placed in different administrative domains, run different operating systems, and so forth,” they wrote.
The honeychecker can also be designed so it doesn’t directly interface with the Internet, which would also reduce an attacker’s ability to hack it, Barrett noted.
Not a total fix
Use of honeywords isn’t going to prevent hackers from stealing password databases and cracking their secrets, Juels and Rivest acknowledge.
“However,” they add in their paper, “the big difference when honeywords are used is that a successful brute-force password break does not give the adversary confidence that he can log in successfully and undetected.”
“The use of a honeychecker,” the authors said, “thus forces an adversary to either risk logging in with a large chance of causing the detection of the compromise of the password-hash… or else to attempt compromising the honeychecker as well.”
The researchers admit that honeywords aren’t a wholly satisfying solution to user authentication on the Net because the scheme contains many of the known problems with passwords and “something-you-know” authentication generally.
“Eventually,” they wrote, “passwords should be supplemented with stronger and more convenient authentication methods…or give way to better authentication methods completely.”