Coherent attacks are stronger than collective attacks on DIQKD with random postselection

TL;DR

This paper demonstrates that in device-independent quantum key distribution with random postselection, coherent attacks can outperform collective attacks, challenging previous security assumptions and highlighting the need for more robust security proofs.

Contribution

The paper provides the first explicit attack showing that coherent attacks are more powerful than collective attacks against this protocol.

Findings

Coherent attacks can break the protocol's security.

Previous proofs only considered collective attacks.

Security against coherent attacks remains unproven.

Abstract

In a recent paper [Phys. Rev. Lett. 129, 050502 (2022)], the authors report on the implementation of a device-independent QKD protocol with random postselection, which was originally proposed in [Phys. Rev. Lett. 128, 110506 (2022)]. Both works only provide a security proof against collective attacks, leaving open the question whether the protocol is secure against coherent attacks. In this note, we report on an attack on this protocol that demonstrates that coherent attacks are, in fact, stronger than collective attacks.