Security boundaries of semi-quantum private comparison with self-reported third-party verification

Abstract

Abstract Semi-quantum private comparison (SQPC) security depends not only on quantum resources, but also on how verification evidence is generated and trusted. We examine the single-particle SQPC protocol of Ye et al as a representative case. Its source-defined semi-honest third party (TP) may perform possible attacks subject to non-collusion, while the prescribed TP-honesty checks rely on measurement evidence produced by the TP itself. We therefore investigate a reporting-trust boundary: an active-TP interpretation in which a malicious TP may falsify its own authenticated measurement reports or checking decisions. Under this adversarial self-reporting model, causally valid fake-state and intercept–measure–resend attacks recover the Alice–Bob intermediate key K A B and hence both individual private inputs without using future Alice/Bob operation information, while preserving the intended comparison relation. The prescribed checking procedure does not reject these attacks under the stated self-reporting interpretation. A participant-side replacement attack provides a contrast: it can compromise confidentiality on surviving executions, but the original first check detects the full-replacement strategy with high probability. The analysis distinguishes confidentiality, detectability, and comparison correctness, identifies four structural screening factors, and uses secret separation as a reference enhancement illustrating the independent secret resource needed to protect individual inputs. Protocol-level Monte Carlo results agree with the analytical predictions. Under adversarial self-reporting, both TP-side attacks achieve unit per-bit private-input reconstruction rates while maintaining unit comparison-correctness rates and producing zero rejection by the prescribed checking procedure. In contrast, the participant-side full-replacement strategy is detected with probability approximately 0.984 at N = 16 . With secret separation, individual per-bit reconstruction returns to approximately the 0.5 random-guessing baseline while comparison correctness remains unchanged.