Can Agents Secure Hardware? Evaluating Agentic LLM-Driven Obfuscation for IP Protection
Sujan Ghimire, Parsa Mirfasihi, Muhtasim Alam Chowdhury, Veeramani Pugazhenthi, Harish Kumar Dharavath + 4 more
TLDR
Evaluates an LLM-driven agentic framework for automating hardware netlist obfuscation to protect IP, showing both its potential and current limitations.
Key contributions
- Introduces an LLM-driven agentic framework for automated hardware netlist obfuscation to protect IP.
- Framework uses retrieval-grounded planning, structured lock-plan generation, and multi-stage decomposition.
- Evaluated on ISCAS-85 benchmarks, demonstrating functional correctness and output corruption with wrong keys.
- Highlights potential of LLM-driven obfuscation but notes its current vulnerability to SAT attacks.
Why it matters
This paper addresses critical hardware IP protection in global supply chains by proposing an automated LLM-driven obfuscation framework. It demonstrates the potential of agentic LLMs for hardware security, while highlighting current vulnerabilities. This guides future research towards more robust, AI-driven protection methods.
Original Abstract
The globalization of integrated circuit (IC) design and manufacturing has increased the exposure of hardware intellectual property (IP) to untrusted stages of the supply chain, raising concerns about reverse engineering, piracy, tampering, and overbuilding. Hardware netlist obfuscation is a promising countermeasure, but automating the generation of functionally correct and security-relevant obfuscated circuits remains challenging, particularly for benchmark-scale designs. This paper presents an agentic, large language model (LLM)-driven framework for automated hardware netlist obfuscation. The proposed framework combines retrieval-grounded planning, structured lock-plan generation, deterministic netlist compilation, functional verification, and SAT-based security evaluation. Rather than a single prompt-to-output generation step, the framework decomposes the task into specialized stages for circuit analysis, synthesis, verification, and attack evaluation. We evaluate the framework on ISCAS-85 benchmarks using functional equivalence checking and SAT-based attacks. Results show that the framework generates correct locked netlists while introducing measurable output corruption under incorrect keys, while SAT attacks remain effective. These findings highlight both the potential and current limitations of agentic LLM-driven obfuscation.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.