LLA: Enhancing Security and Privacy for Generative Models with Logic-Locked Accelerators

TL;DR

This paper presents LLA, a hardware-software co-designed scheme that enhances security and privacy of generative AI models by embedding keys and locking mechanisms into accelerators, effectively preventing IP theft and information leakage.

Contribution

It introduces a novel logic-locked accelerator design that combines software obfuscation techniques with hardware security modules to protect generative models against supply chain threats.

Findings

LLA withstands oracle-guided key optimization attacks.

Minimal overhead of less than 0.1% for 7,168 key bits.

Effective protection against model theft, corruption, and leakage.

Abstract

We introduce LLA, an effective intellectual property (IP) protection scheme for generative AI models. LLA leverages the synergy between hardware and software to defend against various supply chain threats, including model theft, model corruption, and information leakage. On the software side, it embeds key bits into neurons that can trigger outliers to degrade performance and applies invariance transformations to obscure the key values. On the hardware side, it integrates a lightweight locking module into the AI accelerator while maintaining compatibility with various dataflow patterns and toolchains. An accelerator with a pre-stored secret key acts as a license to access the model services provided by the IP owner. The evaluation results show that LLA can withstand a broad range of oracle-guided key optimization attacks, while incurring a minimal computational overhead of less than 0.1% for 7,168 key bits.