InfoResearchPreprintLLM-specific
GuardAgent: A Memory-Augmented Agentic Framework for Adaptive Detection of Adversarial Promotions in LLM-Based Recommendation
- Published
- Record updated
Summary
This paper presents the first systematic study on detecting promotion-driven attacks against LLM-based recommender systems, where textual content becomes the main medium for generating recommendations. The authors propose GuardAgent, an LLM-powered agentic detection framework that runs a loop of Perception, Action, Reflection and Memory. The authors report state-of-the-art generalization on real-world datasets, though the source gives no specific figures.
Topics
Related items
- HighCVE-2026-77283: obsidian-sync-mcp HTTP /mcp endpoint DNS rebinding and cross-origin vault accessSimilar attack · NVD/CVE Database
- HighCVE-2026-75612: yt-mcp arbitrary file read through add_attachment file_pathSimilar attack · NVD/CVE Database
- MediumCVE-2026-75518: DOMShell MCP server eval tool bypasses write permission checksSimilar attack · NVD/CVE Database
- HighCVE-2026-62964: Gortex arbitrary file write through MCP file toolsSimilar attack · NVD/CVE Database
- HighCVE-2026-127222: SGLang missing authorization on POST /load_lora_adapter_from_tensorsSimilar attack · NVD/CVE Database