Quantum Security Technologies

The Quantum Security Technologies (QST) research group explores the intersection of quantum computing, artificial intelligence (AI), optimization and IT security. As part of Fraunhofer AISEC’s quantum mission “Security despite, for and with Quantum Computing”, the QST group focuses on the latter two pillars — security for and with quantum computers — while the first pillar, security despite quantum computing, is addressed by Fraunhofer AISEC’s Center of post-quantum cryptography excellence.

Quantum-Assisted Methods for Improving IT Security

Fraunhofer AISEC explores how quantum computing can enhance traditional cybersecurity measures. Our work includes developing quantum-assisted AI algorithms for anomaly detection, threat identification and system verification. Using quantum machine learning (QML, the combination of quantum computing and machine learning) and quantum optimization, we aim to reduce computational complexity and improve detection accuracy — even in challenging scenarios. These include high-dimensional data, where a large number of variables must be analyzed at once, or limited labeled datasets, where only a small amount of data has been pre-classified by experts. To ensure practical relevance, our methods are validated using both classical simulations and real quantum hardware.

Security for Quantum Computers and Quantum AI

As quantum systems and quantum-enhanced AI become increasingly prevalent, ensuring their security is crucial. Fraunhofer AISEC analyzes vulnerabilities unique to quantum computing, including vulnerability to quantum-specific adversarial attacks, leakage through quantum side channels and robustness of parameterized quantum circuits. We also design defense mechanisms to protect QML models against perturbations and evaluate the resilience of quantum cryptographic protocols and hardware architectures. Our work supports secure deployment of quantum-enhanced AI in real-world environments.

Fraunhofer AISEC is a national leader in the field of hardening and robustness analysis of AI methods, including the integration of AI with quantum computing. Through high-profile publications in international conferences and close cooperation with our industrial partners, the Quantum Security Technologies research group contributes cutting-edge insights at the intersection of quantum computing and IT security.

We offer both theoretical foundations and practical implementations of quantum-enhanced and secure techniques for future-proof quantum computing infrastructures.

The Quantum Security Technologies group has in-depth expertise in the following areas:

• Anomaly and fraud detection using quantum machine learning
• Adversarial and robust quantum machine learning
• Attack vectors and defenses on quantum computing
• QC-assisted optimization for verification problems

Design and Implementation of QC-assisted Security Solutions

• Design and implementation of proof-of-concept implementation for typical security use-cases such as anomaly or fraud detection
• Noisy intermediate scale quantum (NISQ) or fault-tolerant quantum computing (FTQC) approaches
• Dequantization and quantum-inspired algorithms
• Implementation, evaluation and benchmarking on simulators and real quantum hardware
• Analysis of potential quantum advantages and hardware requirements

Evaluation of QC-assisted Solutions 

• Benchmarking quantum models against classical baselines with a focus on performance, robustness, uncertainty quantification, and interpretability
• Vulnerability analysis of quantum models in adversarial and non-stationary environments
• Development of defensive techniques for quantum models considering the full QC technology stack

Consulting and Training

• Consulting for QC use-cases in IT security and beyond
• Seminars and training courses on QC for IT security

We develop QML-based approaches to identify anomalous behavior in cybersecurity contexts, such as network traffic and cloud monitoring and system log analysis. Leveraging quantum kernels and variational circuits, our models are designed to generalize from a few examples and handle complex, high-dimensional data distributions.

-> Read the blog entry: Anomaly Detection with Quantum Machine Learning: Identifying Cybersecurity Issues in Datasets

We analyze vulnerabilities of QML models in adversarial settings by developing attacks tailored to quantum circuits and designing robust training strategies to counter them. Our research includes studies of parameter sensitivity, trade-offs between circuit depth and robustness, and the development of provable guarantees under both noise and adversarial perturbations.

-> Read the blog entry: Quantum and Classical AI Security: How to Build Robust Models Against Adversarial Attacks

We explore the application of quantum algorithms such as QAOA and VQE to assist in the verification of software and system properties, including bug detection, constraint satisfaction, and symbolic execution. These methods aim to enhance the efficiency and to expand the coverage of classical verification pipelines.