Can LLMs Fool Graph Learning? Exploring Universal Adversarial Attacks on Text-Attributed Graphs

• URL: http://arxiv.org/abs/2603.21155v1
• Date: Sun, 22 Mar 2026 10:15:22 GMT
• Title: Can LLMs Fool Graph Learning? Exploring Universal Adversarial Attacks on Text-Attributed Graphs
• Abstract summary: Text-attributed graphs (TAGs) enhance graph learning by integrating rich textual semantics and topological context for each node.<n>Recent advances leverage diverse backbones, such as graph neural networks (GNNs) and pre-trained language models (PLMs) to capture both structural and textual information in TAGs.<n>This diversity raises a key question: How can we design universal adversarial attacks that generalize across architectures to assess the security of TAG models?<n>We propose BadGraph, a novel attack framework that deeply elicits large language models (LLMs) understanding of general graph knowledge to jointly perturb both node topology and textual semantics
• Score: 22.41904611417485
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Text-attributed graphs (TAGs) enhance graph learning by integrating rich textual semantics and topological context for each node. While boosting expressiveness, they also expose new vulnerabilities in graph learning through text-based adversarial surfaces. Recent advances leverage diverse backbones, such as graph neural networks (GNNs) and pre-trained language models (PLMs), to capture both structural and textual information in TAGs. This diversity raises a key question: How can we design universal adversarial attacks that generalize across architectures to assess the security of TAG models? The challenge arises from the stark contrast in how different backbones-GNNs and PLMs-perceive and encode graph patterns, coupled with the fact that many PLMs are only accessible via APIs, limiting attacks to black-box settings. To address this, we propose BadGraph, a novel attack framework that deeply elicits large language models (LLMs) understanding of general graph knowledge to jointly perturb both node topology and textual semantics. Specifically, we design a target influencer retrieval module that leverages graph priors to construct cross-modally aligned attack shortcuts, thereby enabling efficient LLM-based perturbation reasoning. Experiments show that BadGraph achieves universal and effective attacks across GNN- and LLM-based reasoners, with up to a 76.3% performance drop, while theoretical and empirical analyses confirm its stealthy yet interpretable nature.

Related papers

• Cross-Paradigm Graph Backdoor Attacks with Promptable Subgraph Triggers [49.77729302007601]
  Graph Neural Networks (GNNs) are vulnerable to backdoor attacks, where adversaries implant malicious triggers to manipulate model predictions.<n>Existing trigger generators are often simplistic in structure and overly reliant on specific features, confining them to a single graph learning paradigm.<n>We propose Cross-Paradigm Graph Backdoor Attacks with Promptable Subgraph Triggers(CP-GBA), a new transferable graph backdoor attack.
  arXiv  Detail & Related papers  (2025-10-26T07:10:07Z)
• Unveiling the Vulnerability of Graph-LLMs: An Interpretable Multi-Dimensional Adversarial Attack on TAGs [35.900360659024585]
  Interpretable Multi-Dimensional Graph Attack (IMDGA) is a novel human-centric adversarial attack framework for Graph-LLMs.<n>IMDGA demonstrates superior interpretability, attack effectiveness, stealthiness, and robustness compared to existing methods.<n>This work uncovers a previously underexplored semantic dimension of vulnerability in Graph-LLMs, offering valuable insights for improving their resilience.
  arXiv  Detail & Related papers  (2025-10-14T07:36:07Z)
• GILT: An LLM-Free, Tuning-Free Graph Foundational Model for In-Context Learning [50.40400074353263]
  Graph Neural Networks (GNNs) are powerful tools for precessing relational data but often struggle to generalize to unseen graphs.<n>We introduce textbfGraph textbfIn-context textbfL textbfTransformer (GILT), a framework built on an LLM-free and tuning-free architecture.
  arXiv  Detail & Related papers  (2025-10-06T08:09:15Z)
• TrustGLM: Evaluating the Robustness of GraphLLMs Against Prompt, Text, and Structure Attacks [3.3238054848751535]
  We introduce TrustGLM, a comprehensive study evaluating the vulnerability of GraphLLMs to adversarial attacks across three dimensions: text, graph structure, and prompt manipulations.<n>Our findings reveal that GraphLLMs are highly susceptible to text attacks that merely replace a few semantically similar words in a node's textual attribute.<n>We also find that standard graph structure attack methods can significantly degrade model performance, while random shuffling of the candidate label set in prompt templates leads to substantial performance drops.
  arXiv  Detail & Related papers  (2025-06-13T14:48:01Z)
• LLM as GNN: Graph Vocabulary Learning for Text-Attributed Graph Foundation Models [87.68057302738457]
  Text-Attributed Graphs (TAGs) are ubiquitous in real-world scenarios.<n>Despite large efforts to integrate Large Language Models (LLMs) and Graph Neural Networks (GNNs) for TAGs, existing approaches suffer from decoupled architectures.<n>We propose PromptGFM, a versatile GFM for TAGs grounded in graph vocabulary learning.
  arXiv  Detail & Related papers  (2025-03-05T09:45:22Z)
• Can LLMs Convert Graphs to Text-Attributed Graphs? [35.53046810556242]
  We propose Topology-Aware Node description Synthesis (TANS) to convert existing graphs into text-attributed graphs.<n>We evaluate our TANS on text-rich, text-limited, and text-free graphs, demonstrating its applicability.
  arXiv  Detail & Related papers  (2024-12-13T13:32:59Z)
• Can Graph Neural Networks Learn Language with Extremely Weak Text Supervision? [62.12375949429938]
  We propose a multi-modal prompt learning paradigm to adapt pre-trained Graph Neural Networks to downstream tasks and data.<n>Our new paradigm embeds the graphs directly in the same space as the Large Language Models (LLMs) by learning both graph prompts and text prompts simultaneously.<n>We build the first CLIP-style zero-shot classification prototype that can generalize GNNs to unseen classes with extremely weak text supervision.
  arXiv  Detail & Related papers  (2024-12-11T08:03:35Z)
• Parameter-Efficient Tuning Large Language Models for Graph Representation Learning [62.26278815157628]
  We introduce Graph-aware. Efficient Fine-Tuning - GPEFT, a novel approach for efficient graph representation learning. We use a graph neural network (GNN) to encode structural information from neighboring nodes into a graph prompt. We validate our approach through comprehensive experiments conducted on 8 different text-rich graphs, observing an average improvement of 2% in hit@1 and Mean Reciprocal Rank (MRR) in link prediction evaluations.
  arXiv  Detail & Related papers  (2024-04-28T18:36:59Z)
• G-Retriever: Retrieval-Augmented Generation for Textual Graph Understanding and Question Answering [61.93058781222079]
  We develop a flexible question-answering framework targeting real-world textual graphs. We introduce the first retrieval-augmented generation (RAG) approach for general textual graphs. G-Retriever performs RAG over a graph by formulating this task as a Prize-Collecting Steiner Tree optimization problem.
  arXiv  Detail & Related papers  (2024-02-12T13:13:04Z)
• GraphGPT: Graph Instruction Tuning for Large Language Models [27.036935149004726]
  Graph Neural Networks (GNNs) have evolved to understand graph structures. To enhance robustness, self-supervised learning (SSL) has become a vital tool for data augmentation. Our research tackles this by advancing graph model generalization in zero-shot learning environments.
  arXiv  Detail & Related papers  (2023-10-19T06:17:46Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.