SQUASH: A SWAP-Based Quantum Attack to Sabotage Hybrid Quantum Neural Networks

2025-06-30

AI Summary

This work exposes a critical hardware-level security vulnerability in hybrid quantum neural networks (HQNNs): we propose SQUASH—the first stealthy attack targeting the variational quantum circuit structure. SQUASH strategically inserts SWAP gates into the victim HQNN’s quantum circuit, inducing qubit misalignment and state evolution distortion—without requiring access to training data, without modifying the input state, and without introducing observable perturbations. It supports both untargeted and targeted attacks, reducing classification accuracy by 74.08% and target-class recognition rate by 79.78% in noiseless settings. Crucially, this work pioneers the integration of circuit-level structural manipulation into quantum machine learning security research. By demonstrating that adversarial manipulation at the quantum gate level can severely compromise model integrity, it underscores the urgent need for hardware-aware security design in practical HQNN deployment.

Technology Category

• Machine Learning: Quantum Machine Learning
• Computer Vision: Adversarial Attacks & Robustness
• Humans and AI: Human-in-the-loop Machine Learning

Application Category

• Security and Privacy: Large-scale security measurements
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

We propose a circuit-level attack, SQUASH, a SWAP-Based Quantum Attack to sabotage Hybrid Quantum Neural Networks (HQNNs) for classification tasks. SQUASH is executed by inserting SWAP gate(s) into the variational quantum circuit of the victim HQNN. Unlike conventional noise-based or adversarial input attacks, SQUASH directly manipulates the circuit structure, leading to qubit misalignment and disrupting quantum state evolution. This attack is highly stealthy, as it does not require access to training data or introduce detectable perturbations in input states. Our results demonstrate that SQUASH significantly degrades classification performance, with untargeted SWAP attacks reducing accuracy by up to 74.08% and targeted SWAP attacks reducing target class accuracy by up to 79.78%. These findings reveal a critical vulnerability in HQNN implementations, underscoring the need for more resilient architectures against circuit-level adversarial interventions.

Problem

• Proposes SQUASH attack to sabotage Hybrid Quantum Neural Networks
• Inserts SWAP gates to disrupt quantum state evolution
• Reveals critical vulnerability in HQNN implementations

Innovation

• SWAP gate insertion in quantum circuits
• Direct manipulation of circuit structure
• Stealthy attack without data access