Joint-GCG: Unified Gradient-Based Poisoning Attacks on Retrieval-Augmented Generation Systems
Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (42) · 35793-35801
DOI: 10.1609/aaai.v40i42.40893 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Haowei Wang und 6 weiteren Autor:innen. Die Arbeit wurde bisher 4-mal zitiert (Crossref).
Abstract (Originalsprache)
Retrieval-Augmented Generation (RAG) systems enhance Large Language Models (LLMs) by retrieving relevant documents from external corpora before generating responses. This approach significantly expands LLM capabilities by leveraging vast, up-to-date external knowledge. However, this reliance on external knowledge makes RAG systems vulnerable to corpus poisoning attacks that manipulate generated outputs via poisoned document injection. Existing poisoning attack strategies typically treat the retrieval and generation stages as disjointed, limiting their effectiveness. We propose Joint-GCG, the first framework to unify gradient-based attacks across both retriever and generator models through three innovations: (1) Cross-Vocabulary Projection for aligning embedding spaces, (2) Gradient Tokenization Alignment for synchronizing token-level gradient signals, and (3) Adaptive Weighted Fusion for dynamically balancing attacking objectives. Evaluations demonstrate that Joint-GCG achieves at most 25% and an average of 5% higher attack success rate than previous methods across multiple retrievers and generators. While optimized under a white-box assumption, the generated poisons show unprecedented transferability to unseen models. Joint-GCG's innovative unification of gradient-based attacks across retrieval and generation stages fundamentally reshapes our understanding of vulnerabilities within RAG systems.
Zitieren
APA
Wang, H., Zhang, R., Wang, J., Li, M., Huang, Y., Wang, D., & Wang, Q. (2026). Joint-GCG: Unified Gradient-Based Poisoning Attacks on Retrieval-Augmented Generation Systems. Proceedings of the AAAI Conference on Artificial Intelligence, 40(42), 35793-35801. https://doi.org/10.1609/aaai.v40i42.40893
Vancouver
Wang H, Zhang R, Wang J, Li M, Huang Y, Wang D, et al. Joint-GCG: Unified Gradient-Based Poisoning Attacks on Retrieval-Augmented Generation Systems. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(42):35793-35801. doi:10.1609/aaai.v40i42.40893
BibTeX
@article{Wang2026,
author = {Wang, Haowei and Zhang, Rupeng and Wang, Junjie and Li, Mingyang and Huang, Yuekai and Wang, Dandan and Wang, Qing},
title = {Joint-GCG: Unified Gradient-Based Poisoning Attacks on Retrieval-Augmented Generation Systems},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2026},
volume = {40},
number = {42},
pages = {35793--35801},
doi = {10.1609/aaai.v40i42.40893}
}
EUVIMED