Archer Materials QNN Matches Classical AI in Quantum Fraud Detection Milestone
Archer Materials’ QNN model matches classical fraud detection in milestone quantum test
Archer Materials (ASX: AXE) has completed the simulation and benchmarking stage of its quantum machine learning (QML) fraud detection project, following the dataset preparation milestone announced in March 2026. The headline result is striking: a quantum neural network (QNN) correctly identified 118 fraudulent transactions with just one false positive, a performance level equivalent to the best classical models tested.
The QNN was also executed on IQM Garnet, a commercial 20-qubit superconducting quantum computer accessed through AWS Braket, successfully detecting 18 of 19 fraudulent transactions on real hardware.
Dr Simon Ruffell, CEO
“These results have demonstrated that QML approaches can deliver strong fraud detection performance while operating within the constraints of current quantum computing systems.”
When big ASX news breaks, our subscribers know first
What the results actually show
Simulator performance
The QNN was tested on a publicly available financial fraud dataset containing more than 280,000 transaction records. The dataset was processed using dimensionality reduction and data balancing techniques to enable operation within current quantum computing constraints.
Key simulator results:
- 118 fraudulent transactions correctly identified
- 30 fraudulent transactions missed
- Only 1 false positive generated
- Performance equivalent to best classical benchmarking models
- Model remained stable under simulated quantum noise at low noise levels, with only minor performance degradation at moderate noise levels; material degradation observed only at higher noise levels
Real hardware validation
Beyond simulation, the QNN was executed on IQM Garnet via AWS Braket, marking a transition from controlled simulator testing to real quantum hardware. The hardware run detected 18 of 19 fraudulent transactions in the test set, confirming the model can operate on a commercial superconducting quantum computer.
It is worth noting that higher false-positive rates were observed on hardware compared to the simulator. Archer acknowledges this openly in the announcement, and it represents a useful technical finding about the hardware quality and noise tolerances that practical QML applications may require in the future.
Archer has not yet demonstrated quantum advantage over classical AI approaches. The company frames this explicitly as the next challenge to be addressed, not a present claim.
| Test Environment | Fraudulent Transactions Detected | Missed | False Positives | Status |
|---|---|---|---|---|
| Qubit simulator | 118 | 30 | 1 | Equivalent to best classical models |
| IQM Garnet (real hardware) | 18 of 19 | 1 | Higher than simulator | Validated on commercial quantum hardware |
Why quantum fraud detection matters — and why Archer is building in a high-value space
Fraud detection is one of the more compelling use cases for QML because the operational demands are severe. Banks and payment providers must process enormous volumes of transactions in near real-time, minimising both missed fraud and false alerts. Excessive false positives create costly manual review workflows and poor customer experiences, so precision is as important as recall.
QML is being explored because quantum systems may eventually process complex, high-dimensional data faster than classical computers. This is the “quantum advantage” hypothesis: the idea that quantum architectures could outperform classical AI on specific tasks once hardware matures sufficiently.
Importantly, Archer is not operating in isolation. Quantinuum and HSBC are actively exploring QML for fraud detection, and Intesa Sanpaolo is collaborating with IBM to explore QML for improving fraud detection accuracy and speed. These are major global financial institutions committing real resources to the same application space.
Archer’s research collaboration agreement with the CSIRO forms part of the company’s strategy to investigate practical applications of quantum computing technologies and support future commercialisation opportunities in data-intensive industries.
The next major ASX story will hit our subscribers first
What’s next — full QML prototype targeted by end of 2026
The next phase of the project will involve:
- Testing on larger quantum systems with richer feature representations
- Larger datasets and additional classical benchmarks
- Repeated trials and further hardware validation
- Target: full QML prototype by end of 2026
This phase was conducted on a prepared research dataset under current quantum computing constraints. Any commercial deployment pathway assessment will require further work beyond what has been completed to date.
The completion of this milestone has “substantially reduced technical uncertainty” and established a repeatable benchmarking framework for the next phase, according to the company.
Dr Simon Ruffell, CEO
“Fraud detection is a relevant use case for QML because banks and payment providers must analyse large volumes of transaction data quickly, while reducing both missed fraud and false alerts.”
Don’t Miss the Next ASX Tech Breakthrough
Big News Blast delivers FREE breaking ASX tech news to your inbox within minutes of release, complete with in-depth analysis already done for you. Join 20,000+ investors who stay ahead of market-moving announcements the moment they drop. Click the “Free Alerts” button at StockWire X to start receiving alerts today.