11 papers
The Deliberative Illusion: Diagnosing Factual Attrition and Stance Homogenization in Multi-Agent LLM Deliberation
Herun Wan, Jiaying Wu, Minnan Luo +4
Multi-agent LLM systems often treat consensus as evidence of successful interaction. For deliberative problems, however, reliability depends on whether agents preserve the facts an…
Correct When Paired, Wrong When Split: Decoupling and Editing Modality-Specific Neurons in MLLMs
Tingchao Fu, Wenkai Wang, Fanxiao Li +6
Although Knowledge Editing provides an efficient mechanism for updating the knowledge of Multimodal Large Language Models (MLLMs), we find that current paradigms still suffer from…
Better with Experience: Self-Evolving LLM Agents for Evidence-Grounded Health Community Notes
Zihang Fu, Fanxiao Li, Jianyang Gu +5
Large Language Model (LLM)-augmented Community Notes offer a scalable path for timely, evidence-grounded correction of health misinformation on social platforms. However, they stil…
FlowSteer: Prompt-Only Workflow Steering Exposes Planning-Time Vulnerabilities in Multi-Agent LLM Systems
Fanxiao Li, Jiaying Wu, Tingchao Fu +3
Multi-agent systems (MAS) powered by large language models (LLMs) increasingly adopt planner--executor architectures, where planners convert prompts into subtasks, roles, dependenc…
REFORGE: Multi-modal Attacks Reveal Vulnerable Concept Unlearning in Image Generation Models
Yong Zou, Haoran Li, Fanxiao Li +5
Recent progress in image generation models (IGMs) enables high-fidelity content creation but also amplifies risks, including the reproduction of copyrighted content and the generat…
The Facade of Truth: Uncovering and Mitigating LLM Susceptibility to Deceptive Evidence
Herun Wan, Jiaying Wu, Minnan Luo +3
To reliably assist human decision-making, LLMs must maintain factual internal beliefs against misleading injections. While current models resist explicit misinformation, we uncover…
What's Left Unsaid? Detecting and Correcting Misleading Omissions in Multimodal News Previews
Fanxiao Li, Jiaying Wu, Tingchao Fu +4
Even when factually correct, social-media news previews (image-headline pairs) can induce interpretation drift: by selectively omitting crucial context, they lead readers to form j…
Beyond the Crowd: LLM-Augmented Community Notes for Governing Health Misinformation
Jiaying Wu, Zihang Fu, Haonan Wang +4
Community Notes, the crowd-sourced misinformation governance system on X (formerly Twitter), allows users to flag misleading posts, attach contextual notes, and rate the notes' hel…
Drifting Away from Truth: GenAI-Driven News Diversity Challenges LVLM-Based Misinformation Detection
Fanxiao Li, Jiaying Wu, Tingchao Fu +3
The proliferation of multimodal misinformation poses growing threats to public discourse and societal trust. While Large Vision-Language Models (LVLMs) have enabled recent progress…
CMIE: Combining MLLM Insights with External Evidence for Explainable Out-of-Context Misinformation Detection
Fanxiao Li, Jiaying Wu, Canyuan He +1
Multimodal large language models (MLLMs) have demonstrated impressive capabilities in visual reasoning and text generation. While previous studies have explored the application of…
Seeing Through Deception: Uncovering Misleading Creator Intent in Multimodal News with Vision-Language Models
Jiaying Wu, Fanxiao Li, Zihang Fu +2
The impact of multimodal misinformation arises not only from factual inaccuracies but also from the misleading narratives that creators deliberately embed. Interpreting such creato…