Analysis of A Malicious Deutsch-Jozsa Circuit

Active research into multi-tenancy means multitenant quantum computers could be soon. However, the growth of quantum computing introduces new security risks. One such risk is crosstalk error in multi-tenant superconducting quantum computers, which can be used to inject faults into other users' circuits. Concurrent works in the field have designed malware that exploits this vulnerability. This work demonstrates a malicious Deutsch-Jozsa malware that resists mitigation. This is created by designing a Deutsch-Jozsa circuit that contains a malicious payload. To analyze this malware, we assess both covertness and mitigation resilience. For covertness, we evaluated the impact on the integrity of the Deutsch-Jozsa circuit it hides in. Our findings show that the malware closely mimics the performance of an innocent Deutsch-Jozsa circuit at all optimization levels, which shows success in covertness. To assess mitigation resilience, we use Qiskit's circuit optimization. We found that our malware remains mostly unmitigated by optimization and almost doubles the victim error rate at the highest level of optimization offered by IBM. This shows that hiding a well-designed payload within a wrapping circuit does not reduce the malware's effectiveness, and it even performs better than the non-deceptive variant.