Fuzzing the brain: Automated stress testing for the safety of ML-driven neurostimulation

2025-12-04

AI Summary

Machine learning–driven neural prostheses (e.g., visual prostheses) pose biophysical safety risks—including excessive charge density, transient current surges, and unintended electrode co-activation—arising from direct neural stimulation by model outputs. This paper introduces the first coverage-guided fuzz testing framework tailored to neural stimulation safety violations, integrating black-box input perturbation with real-time monitoring of biophysical constraints to automatically uncover unsafe stimulation patterns. We propose two novel output coverage metrics that formalize safety as a quantifiable, interpretable model property—moving beyond heuristic rules—and enable consistent, architecture- and training-agnostic safety evaluation. Empirically validated on retinal and cortical deep stimulation encoders, our approach reveals diverse, previously undetected safety hazards. The results establish an evidence-based foundation for developing safety benchmarks and regulatory-ready assessment protocols for neural interfaces.

Technology Category

• Natural Language Processing: Safety and Robustness
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Security and Privacy: Large-scale security measurements
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Objective: Machine learning (ML) models are increasingly used to generate electrical stimulation patterns in neuroprosthetic devices such as visual prostheses. While these models promise precise and personalized control, they also introduce new safety risks when model outputs are delivered directly to neural tissue. We propose a systematic, quantitative approach to detect and characterize unsafe stimulation patterns in ML-driven neurostimulation systems. Approach: We adapt an automated software testing technique known as coverage-guided fuzzing to the domain of neural stimulation. Here, fuzzing performs stress testing by perturbing model inputs and tracking whether resulting stimulation violates biophysical limits on charge density, instantaneous current, or electrode co-activation. The framework treats encoders as black boxes and steers exploration with coverage metrics that quantify how broadly test cases span the space of possible outputs and violation types. Main results: Applied to deep stimulus encoders for the retina and cortex, the method systematically reveals diverse stimulation regimes that exceed established safety limits. Two violation-output coverage metrics identify the highest number and diversity of unsafe outputs, enabling interpretable comparisons across architectures and training strategies. Significance: Violation-focused fuzzing reframes safety assessment as an empirical, reproducible process. By transforming safety from a training heuristic into a measurable property of the deployed model, it establishes a foundation for evidence-based benchmarking, regulatory readiness, and ethical assurance in next-generation neural interfaces.

Problem

• Detects unsafe stimulation patterns in ML-driven neurostimulation systems
• Adapts coverage-guided fuzzing to test neural stimulation safety limits
• Transforms safety assessment into an empirical, reproducible process for neural interfaces

Innovation

• Adapts coverage-guided fuzzing to neural stimulation safety testing
• Treats encoders as black boxes using coverage metrics for exploration
• Transforms safety into measurable property for evidence-based benchmarking