Adversarial Attacks Against Automated Fact-Checking: A Survey

2025-09-10

AI Summary

In the context of rampant misinformation, automated fact-checking (AFC) systems are increasingly vulnerable to adversarial attacks, undermining their reliability. This paper systematically classifies adversarial attack vectors targeting AFC into three categories: claim manipulation, evidence perturbation, and claim-evidence pair tampering—the first such taxonomy in the literature. We propose an integrated analytical framework comprising attack generation, evidence-level perturbation, and robustness evaluation, grounded in a comprehensive survey and multi-technique empirical validation—including natural language inference (NLI)-based attacks and multimodal adversarial example generation—to identify critical model vulnerabilities. Concurrently, we synthesize state-of-the-art defense strategies and pinpoint persistent challenges, notably robust training and trustworthy evidence provenance. Our work provides both theoretical foundations and actionable methodologies for enhancing AFC system resilience, thereby advancing the development of trustworthy information verification infrastructures.

Technology Category

• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Multiagent Systems: Adversarial Agents
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Responsible Web: Mitigating misinformation and disinformation

Abstract

In an era where misinformation spreads freely, fact-checking (FC) plays a crucial role in verifying claims and promoting reliable information. While automated fact-checking (AFC) has advanced significantly, existing systems remain vulnerable to adversarial attacks that manipulate or generate claims, evidence, or claim-evidence pairs. These attacks can distort the truth, mislead decision-makers, and ultimately undermine the reliability of FC models. Despite growing research interest in adversarial attacks against AFC systems, a comprehensive, holistic overview of key challenges remains lacking. These challenges include understanding attack strategies, assessing the resilience of current models, and identifying ways to enhance robustness. This survey provides the first in-depth review of adversarial attacks targeting FC, categorizing existing attack methodologies and evaluating their impact on AFC systems. Additionally, we examine recent advancements in adversary-aware defenses and highlight open research questions that require further exploration. Our findings underscore the urgent need for resilient FC frameworks capable of withstanding adversarial manipulations in pursuit of preserving high verification accuracy.

Problem

• Adversarial attacks manipulate automated fact-checking systems
• Existing systems lack resilience against claim and evidence manipulation
• Survey categorizes attack methodologies and evaluates their impacts

Innovation

• Categorizing adversarial attack methodologies against fact-checking systems
• Evaluating impact of attacks on automated fact-checking model performance
• Examining adversary-aware defense mechanisms for enhanced robustness