AI Paper Atlas

Research map / Adversarial robustness and security / Backdoor and poisoning

Fact2Fiction: Targeted Poisoning Attack to Agentic Fact-checking System

Haorui He, Yupeng Li, Bin Benjamin Zhu, Dacheng Wen, Reynold Cheng, Francis C. M. Lau

AAAI 2026 · Adversarial robustness and security › Backdoor and poisoning · approach: Backdoors in large multimodal models

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use Fact2Fiction instead of PoisonedRAGattack on DEFAME, 1% poison · AVeriTeC devASR (%): 33.5 -> 42.4Improved
use Fact2Fiction instead of PoisonedRAGattack on Simple RAG, 1% poison · AVeriTeC devASR (%): 42.4 -> 43.4No clear effect
add answer planning using justifications (ablation)Fact2Fiction on DEFAME, 1% · AVeriTeC devASR (%): 40.9 -> 42.4Improved
add budget planning (ablation)Fact2Fiction on DEFAME, 1% · AVeriTeC devASR (%): 34.6 -> 42.4Improved
add budget planning (ablation)Fact2Fiction on InFact, 8% · AVeriTeC devASR (%): 56.9 -> 59.9Improved
add query planning (ablation)Fact2Fiction on DEFAME, 8% · AVeriTeC devASR (%): 49.4 -> 63.6Improved
add query planning (ablation)Fact2Fiction on Simple RAG, 8% · AVeriTeC devASR (%): 61.5 -> 65.7Improved
target agentic system instead of simple RAGPoisonedRAG, 8% · AVeriTeC devASR (%): 57.4 -> 42.4 (DEFAME), 45.3 (InFact)Hurt
apply paraphrasing defenseFact2Fiction on DEFAME, 1% · AVeriTeC devASR (%): 42.4 -> 39.4Hurt
apply malicious detection defense (k-means)PoisonedRAG vs Fact2Fiction, DEFAME 1% · AVeriTeC devASR (%): PoisonedRAG 33.5 -> 10.0; Fact2Fiction 42.4 -> 26.0Hurt
increase poison rate for Naive/Prompt InjectionNaive attack on DEFAME · AVeriTeC devASR (%): 17.8 -> 18.2 (1% -> 8%)No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy