Mechanism for the prevention of password reuse through Anonymized Hashes

Abstract

Password authentication is an essential and widespread form of user authentication on the Internet with no other authentication system matching its dominance. When a password on one website is breached, if reused, the stolen password can be used to gain access to multiple other authenticated websites. Even amongst technically educated users, the security issues surrounding password reuse are not well understood and restrictive password composition rules have been unsuccessful in reducing password reuse. In response, the US NIST have published standards outlining that, when setting passwords, authentication systems should validate that user passwords have not already been compromised or breached. We propose a mechanism to allows for clients to anonymously validate whether or not a password has been identified in a compromised database, without needing to download the entire database or send their password to a third-party service. A mechanism is proposed whereby password hash data is generalized such that it holds the k-anonymity property. An implementation is constructed to identify to what extent the data should be generalized for it to hold k-anonymity and additionally to group password hashes by their generalized anonymous value. The implementation is run on a database of over 320 million leaked passwords and the results of the anonymization process are considered. Password authentication is an essential and widespread form of user authentication on the Internet with no other authentication system matching its dominance. When a password on one website is breached, if reused, the stolen password can be used to gain access to multiple other authenticated websites. Even amongst technically educated users, the security issues surrounding password reuse are not well understood and restrictive password composition rules have been unsuccessful in reducing password reuse. In response, the US NIST have published standards outlining that, when setting passwords, authentication systems should validate that user passwords have not already been compromised or breached. We propose a mechanism to allows for clients to anonymously validate whether or not a password has been identified in a compromised database, without needing to download the entire database or send their password to a third-party service. A mechanism is proposed whereby password hash data is generalized such that it holds the k-anonymity property. An implementation is constructed to identify to what extent the data should be generalized for it to hold k-anonymity and additionally to group password hashes by their generalized anonymous value. The implementation is run on a database of over 320 million leaked passwords and the results of the anonymization process are considered.

Practical Hash-based Anonymity for MAC Addresses

“…(Li et al, 2019) vides an overview of protocols for compromised credential checking (C3) services and a mechanism for empirically evaluating such protocols. Such protocols are based off the HIBP protocol whereby passwords are bucketed based on a hash prefix, allowing for kanonymous verification of a breached password as implemented in (Ali, 2017). This bucket is then served to the user to identify if the password is breached or not.…”

Cross Hashing: Anonymizing encounters in Decentralised Contact Tracing Protocols

“…The key issue to solve here is allowing for users to compute the Rolling Proximity Identifiers of users they may have come into contact with without leaking the information of all affected users. Developers of contact tracing apps [8] have sought guidance on implementing my original k-Anonymity communication protocol [9] to mitigate this effect (a protocol originally devised for compromised credential checking).…”

“…We seek to allow a user to query CCIs whilst minimising the privacy loss both on the server and for the client. Fortunately, this is a well explored problem in [9], [11], [12].…”

“…When a user's Temporary Exposure Key is uploaded to a server, the CCIs are computed for the period of exposure. They are then stored in k-Anonymous buckets described in [9]. The client simply needs to truncate the CCI hash to a desired number of bits such to grant anonymity (we provide both formal definitions and empirical analysis for applying k-Anonymity to hash based data sets in [14]).…”

“…During a k-Anonymous search protocol, this means the client discloses a partial hash as the search partition and then both parties are able to determine if the content of the clients search is in that partition without needing to disclosure any further information. Unlike the approach detailed in [9], this provides further cryptographic shielding of the hashes held by the server.…”