Covert Quantum Computing - a new quantum security paradigm

The talk starts from the expected move to multi-tenant quantum computing, where several users share one quantum processor to avoid idle hardware. It defines covert quantum computing as a guarantee that one user cannot detect whether another is running a computation. Covertness matters because a neighbor who can see activity could learn about an algorithm or its results, correlate usage spikes with insider news, or time a fault injection against someone else's job.

The speaker outlines the quantum software and control stack, then focuses on the qubits themselves. Two-qubit gates leak noise onto neighboring qubits, so a user holding nearby qubits can induce bit flips or detect activity. Prior work on this crosstalk was summarized, including attacks that corrupt a victim's circuit. The experiments used Ramsey interferometry on IQM and IBM superconducting hardware. They found the expected nearest-neighbor crosstalk, plus unexpected long-range effects several hops away, which the speaker suspects come from microwave pulses leaking onto nearby control lines.

The practical takeaway is that these experiments can be reproduced by anyone with a free quantum cloud account, and no quantum background is needed. Suggested next steps include compilers that optimize for covertness, alternative detection methods, and comparing ion-trap and neutral-atom machines. The speaker also notes that covertness applies only at the physical-qubit level, not to error-corrected logical qubits.