International Association
for Cryptologic Research

With the advent of Malicious (Peyrin and Wang, Crypto'20), the question of a cipher with an intentional weakness which is only known to its designer has gained its momentum. In their work, the authors discuss how an otherwise secure cipher can be broken by its designer with the help of a secret backdoor (which is not known to the user/attacker). The contribution of Malicious is to propose a cipher-level construction with a backdoor, where it is computationally infeasible to retrieve the backdoor entry despite knowing how the mechanism works.

In this work, we revisit the work by Peyrin and Wang in a greater depth. We discuss the relevant aspects with more clarity, thereby addressing some of the important issues connected to a backdoor construction. The main contribution, however, comes as a new proof-of-concept block cipher with an innate backdoor, named ZUGZWANG. Unlike Malicious, which needs new/experimental concepts like partially non-linear layer; our cipher entirely relies on concepts which are well-established for decades (such as, using an one-way function as a Feistel cipher's state-update), and also offers quite a few advantages over Malicious (easy to visualise, succeeds with probability 1, and so on). Having known the secret backdoor entry, one can recover the secret key with only 1 plaintext query to our cipher; but it is secure otherwise. As the icing on the cake, we show the provable security claims for our cipher.