Reproduce “Micro-Collaborative Poisoning: A Distributed Attack on RAG Systems” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Pedro Pereira, Eva Maia, Isabel Praça
The paper introduces Micro-Collaborative Poisoning, an attack that distributes one false target across five plausible documents with different rhetorical roles, then evaluates how retrieval and database coverage affect RAG answers. Across 108 configurations, at top-k=5, ASR for openai-gpt-oss-120b was 0.722 with both databases poisoned versus 0.630 with one. Findings are bounded to 100 aligned question pairs.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.