TrustErase: Auditable Instant Machine Unlearning with Passport-Embedded Representations

2026-06-15

AI Summary

Existing machine unlearning methods often suffer from insufficient verifiability and high computational overhead, making them ill-suited for privacy compliance requirements. This work proposes a verifiable, on-demand unlearning framework that operates without access to the original training data. By embedding cryptographic “passport” keys into parameter-efficient adaptation layers and leveraging singular value decomposition, the method achieves modular, auditable unlearning without retraining or data access for the first time. It enables transparent deletion of specific classes or entire datasets while maintaining model utility. Experiments on MNIST, CIFAR-10, and CIFAR-100 demonstrate that the approach matches or exceeds the performance of state-of-the-art methods such as DELETE, L2UL, and Boundary Shrink, significantly improving both unlearning efficiency and compliance credibility.

Abstract

The demand for privacy-compliant AI has amplified the need for machine unlearning; yet, existing retraining or distillation-based methods remain unverifiable and computationally costly. We introduce TrustErase, a verifiable, data-free unlearning framework leveraging passport-embedded representations for instant, modular, and auditable forgetting. By treating passports as cryptographic keys within parameter-efficient adaptation layers, TrustErase enables the removal of specific classes or datasets through simple deactivation, without retraining, fine-tuning, or access to the original data. A singular value based decomposition conceals passports within model weights, ensuring that unlearning actions remain transparent and provably compliant. Evaluations on MNIST, CIFAR10 and CIFAR100 show that TrustErase matches or exceeds state-of-the-art benchmarks such as DELETE, L2UL, and Boundary Shrink, while operating in a strictly data-free regime. Ultimately, TrustErase establishes a new paradigm for trustworthy, accountable, and instantly forgettable AI systems.

Problem

• machine unlearning
• privacy compliance
• auditable forgetting
• data-free unlearning
• trustworthy AI

Innovation

• machine unlearning
• passport-embedded representations
• data-free unlearning
• auditable AI
• parameter-efficient adaptation