Watch Out for Your Guidance on Generation! Exploring Conditional Backdoor Attacks against Large Language Models
원문 초록
Mainstream backdoor attacks on large language models (LLMs) typically set a fixed trigger in the input instance and specific responses for triggered queries. However, the fixed trigger setting (e.g., unusual words) may be easily detected by human detection, limiting the effectiveness and practicality in real-world scenarios. To enhance the stealthiness of backdoor activation, we present a new poisoning paradigm against LLMs triggered by specifying generation conditions, which are commonly adopted strategies by users during model inference. The poisoned model performs normally for output under normal/other generation conditions, while becomes harmful for output under target generation conditions. To achieve this objective, we introduce BrieFool, an efficient attack framework. It leverages the characteristics of generation conditions by efficient instruction sampling and poisoning data generation, thereby influencing the behavior of LLMs under target conditions. Our attack can be generally divided into two types with different targets: Safety unalignment attack and Ability degradation attack. Our extensive experiments demonstrate that BrieFool is effective across safety domains and ability domains, achieving higher success rates than baseline methods, with 94.3 % on GPT-3.5-turbo
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
TrojanStego: Your Language Model Can Secretly Be A Steganographic Privacy Leaking Agent
Person detection and re-identification in open-world settings of retail stores and public spaces
MultiSensor-Home: A Wide-area Multi-modal Multi-view Dataset for Action Recognition and Transformer-based Sensor Fusion
Characterizing Photorealism and Artifacts in Diffusion Model-Generated Images
Improving Detection of Person Class Using Dense Pooling
MicroFlow: An Efficient Rust-Based Inference Engine for TinyML
토론
이 논문과 연결된 포럼 글이 아직 없습니다.