What specific vulnerabilities in quantum hardware implementations might attackers actually exploit today?

# Quantum Hardware Vulnerabilities Attackers Can Exploit Today

Several concrete attack vectors exist on current quantum hardware, mostly exploiting the gap between what programmers think is happening and what the physical devices actually do.

Pulse-Level Circuit Attacks

The most direct vulnerability lies in the mismatch between high-level quantum gates and their low-level pulse implementations [AG-2024.06-1682]. When quantum software developers write a custom gate, they specify it at an abstract level. But underneath, that gate executes as a precisely timed sequence of electromagnetic pulses. An attacker with access to the pulse specification can modify those pulses—changing their duration, frequency, phase, or waveform—to corrupt or steal quantum information. The research demonstrates seven attack types on real IBM Quantum hardware, including "qubit plunder" (extracting a qubit's state), "qubit block" (disabling qubits), and timing mismatches that cause wrong answers. Most current quantum software development kits are vulnerable [AG-2024.06-1682].

Crosstalk Exploitation in Shared Quantum Clouds

Crosstalk—unwanted electromagnetic interference between neighboring qubits—is a significant, exploitable source of error on current machines [AG-2024.02-1368]. Because most quantum computers are accessed through shared cloud services, an attacker running a job on nearby qubits can deliberately design their circuit to generate crosstalk noise that sabotages a victim's computation. Research on IBM Quantum hardware shows this is a viable threat today; attackers can intentionally corrupt results [AG-2024.02-1368].

Information Leakage via Readout Crosstalk

In superconducting quantum computers that use frequency-multiplexing to scale up—a standard technique to reduce physical cables—readout crosstalk introduces correlated measurement errors that can leak information about victim qubits [AG-2024.05-2051]. An attacker can predict the quantum state of another user's qubits by exploiting these correlated errors, extracting sensitive data from shared hardware.

State Leakage During Reset

Between quantum algorithm runs, qubits must be reset to their ground state. However, noisy reset operations can leave residual information from the previous computation, which leaks into subsequent runs [AG-2024.01-1977]. An attacker running code after a victim can extract information about the victim's results by observing these leakage patterns.

Current Risk Level

It's important to note that existing quantum computers are not yet mature enough to pose cryptanalytic threats to current encryption schemes [AG-2024.01-2022], and these attacks require direct access to the hardware or the ability to run code in the same shared environment. But for anyone using cloud quantum computers today—whether for algorithm research, optimization, or machine learning—these vulnerabilities are real and exploitable by co-tenants.