AI Backdoor - Papers

1. An LLM-Assisted Easy-to-Trigger Backdoor Attack on Code Completion Models: Injecting Disguised Vulnerabilities against Strong Detection
   • Comments: To appear in USENIX Security ‘24
   • introduce CodeBreaker, a pioneering LLM-assisted backdoor attack framework
   • code completion task
   • https://arxiv.org/abs/2406.06822
   • [Submitted on 10 Jun 2024]
2. BadAgent: Inserting and Activating Backdoor Attacks in LLM Agents
   • Comments: Accepted by ACL 2024
   • propose BadAgent, a robust backdoor attack method for LLM-based intelligent agents
   • LLM agent tasks
   • https://doi.org/10.48550/arXiv.2406.03007
   • [Submitted on 5 Jun 2024]
3. RTL-Breaker: Assessing the Security of LLMs against Backdoor Attacks on HDL Code Generation
   • Comments: Accepted at 2025 Design, Automation & Test in Europe (DATE) Conference
   • propose RTL-Breaker, a novel backdoor attack framework for LLM-based HDL code generation
   • HDL code generation and security risks in hardware systems
   • https://doi.org/10.48550/arXiv.2411.17569
   • [Submitted on 11 Nov 2024]
4. BackdoorLLM: A Comprehensive Benchmark for Backdoor Attacks on Large Language Models
   • Introduces BackdoorLLM, the first comprehensive benchmark for studying backdoor attacks on LLMs
   • propose BackdoorLLM, a benchmark for evaluating diverse backdoor attacks in LLMs
   • backdoor attacks in LLM text generation tasks
   • https://doi.org/10.48550/arXiv.2408.12798
   • [Submitted on 27 Aug 2024]