LLA: Enhancing Security and Privacy for Generative Models with Logic-Locked Accelerators
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911340550946816 |
|---|---|
| author | Li, You Zhao, Guannan Ju, Yuhao He, Yunqi Gu, Jie Zhou, Hai |
| author_facet | Li, You Zhao, Guannan Ju, Yuhao He, Yunqi Gu, Jie Zhou, Hai |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_22307 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | LLA: Enhancing Security and Privacy for Generative Models with Logic-Locked Accelerators Li, You Zhao, Guannan Ju, Yuhao He, Yunqi Gu, Jie Zhou, Hai Cryptography and Security Artificial Intelligence Machine Learning 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. |
| title | LLA: Enhancing Security and Privacy for Generative Models with Logic-Locked Accelerators |
| topic | Cryptography and Security Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2512.22307 |