Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

JSTprove: Pioneering Verifiable AI for a Trustless Future

Jonathan Gold, Tristan Freiberg, Haruna Isah, Shirin Shahabi

TL;DR

JSTprove tackles the challenge of trustworthy AI by providing an end-to-end verifiable inference pipeline that ingests ONNX models, applies quantization, constructs arithmetic circuits, and produces verifiable proofs of inference using a zero-knowledge framework. Built atop the Expander backend, it abstracts cryptographic complexity behind a simple CLI, enabling AI engineers to generate auditable proofs without deep cryptography expertise. The work demonstrates end-to-end verifiable inference for convolutional networks, provides open zkML blueprints for community review, and establishes total circuit cost as a practical predictor of resource usage, guiding future optimizations. Its open-source, auditable approach aims to lower barriers to adoption and to catalyze broader deployment of verifiable AI in privacy- and security-sensitive applications.

Abstract

The integration of machine learning (ML) systems into critical industries such as healthcare, finance, and cybersecurity has transformed decision-making processes, but it also brings new challenges around trust, security, and accountability. As AI systems become more ubiquitous, ensuring the transparency and correctness of AI-driven decisions is crucial, especially when they have direct consequences on privacy, security, or fairness. Verifiable AI, powered by Zero-Knowledge Machine Learning (zkML), offers a robust solution to these challenges. zkML enables the verification of AI model inferences without exposing sensitive data, providing an essential layer of trust and privacy. However, traditional zkML systems typically require deep cryptographic expertise, placing them beyond the reach of most ML engineers. In this paper, we introduce JSTprove, a specialized zkML toolkit, built on Polyhedra Network's Expander backend, to enable AI developers and ML engineers to generate and verify proofs of AI inference. JSTprove provides an end-to-end verifiable AI inference pipeline that hides cryptographic complexity behind a simple command-line interface while exposing auditable artifacts for reproducibility. We present the design, innovations, and real-world use cases of JSTprove as well as our blueprints and tooling to encourage community review and extension. JSTprove therefore serves both as a usable zkML product for current engineering needs and as a reproducible foundation for future research and production deployments of verifiable AI.

JSTprove: Pioneering Verifiable AI for a Trustless Future

TL;DR

JSTprove tackles the challenge of trustworthy AI by providing an end-to-end verifiable inference pipeline that ingests ONNX models, applies quantization, constructs arithmetic circuits, and produces verifiable proofs of inference using a zero-knowledge framework. Built atop the Expander backend, it abstracts cryptographic complexity behind a simple CLI, enabling AI engineers to generate auditable proofs without deep cryptography expertise. The work demonstrates end-to-end verifiable inference for convolutional networks, provides open zkML blueprints for community review, and establishes total circuit cost as a practical predictor of resource usage, guiding future optimizations. Its open-source, auditable approach aims to lower barriers to adoption and to catalyze broader deployment of verifiable AI in privacy- and security-sensitive applications.

Abstract

The integration of machine learning (ML) systems into critical industries such as healthcare, finance, and cybersecurity has transformed decision-making processes, but it also brings new challenges around trust, security, and accountability. As AI systems become more ubiquitous, ensuring the transparency and correctness of AI-driven decisions is crucial, especially when they have direct consequences on privacy, security, or fairness. Verifiable AI, powered by Zero-Knowledge Machine Learning (zkML), offers a robust solution to these challenges. zkML enables the verification of AI model inferences without exposing sensitive data, providing an essential layer of trust and privacy. However, traditional zkML systems typically require deep cryptographic expertise, placing them beyond the reach of most ML engineers. In this paper, we introduce JSTprove, a specialized zkML toolkit, built on Polyhedra Network's Expander backend, to enable AI developers and ML engineers to generate and verify proofs of AI inference. JSTprove provides an end-to-end verifiable AI inference pipeline that hides cryptographic complexity behind a simple command-line interface while exposing auditable artifacts for reproducibility. We present the design, innovations, and real-world use cases of JSTprove as well as our blueprints and tooling to encourage community review and extension. JSTprove therefore serves both as a usable zkML product for current engineering needs and as a reproducible foundation for future research and production deployments of verifiable AI.
Paper Structure (16 sections, 10 equations, 8 figures, 4 tables)

This paper contains 16 sections, 10 equations, 8 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Background and zkML Landscape
  3. ZKP Protocols
  4. zkML Frameworks
  5. System Overview
  6. Circuit Design
  7. Quantization Strategy
  8. Benchmarking
  9. Depth Sweep
  10. Breadth Sweep
  11. Artifact Sizes
  12. Summary
  13. Transparency and Auditability
  14. Conclusion
  15. Acknowledgments
  16. ...and 1 more sections

Figures (8)

  • Figure 1: JSTprove pipeline
  • Figure 2: CNN structure for depth sweep
  • Figure 3: Depth sweep: runtime vs. ECC total cost. The bottom panel aggregates non-compile phases.
  • Figure 4: Depth sweep: peak memory vs. ECC total cost. The bottom panel aggregates non-compile phases.
  • Figure 5: Depth sweep: runtime vs. parameter count. The bottom panel aggregates non-compile phases.
  • ...and 3 more figures

Theorems & Definitions (2)

  • Remark A.1
  • Remark A.2