On the Feasibility of Unclonable Encryption, and More

Unclonable encryption, first introduced by Broadbent and Lord (TQC'20), is a one-time encryption scheme with the following security guarantee: any non-local adversary (𝒜, ℬ, 𝒞) cannot simultaneously distinguish encryptions of two equal length messages. This notion is termed as unclonable indistinguishability. Prior works focused on achieving a weaker notion of unclonable encryption, where we required that any non-local adversary (𝒜, ℬ, 𝒞) cannot simultaneously recover the entire message 𝑚. Seemingly innocuous, understanding the feasibility of encryption schemes satisfying unclonable indistinguishability (even for 1-bit messages) has remained elusive.

We make progress towards establishing the feasibility of unclonable encryption.

• We show that encryption schemes satisfying unclonable indistinguishability exist unconditionally in the quantum random oracle model.

• Towards understanding the necessity of oracles, we present a negative result stipulating that a large class of encryption schemes cannot satisfy unclonable indistinguishability.

• Finally, we also establish the feasibility of another closely related primitive: copy-protection for single-bit output point functions. Prior works only established the feasibility of copyprotection for multi-bit output point functions or they achieved constant security error for single-bit output point functions.