VirusCascade: Hijacking Collaborative Reflection in LLM-Powered Recommender Agents
Factual evidence
What the source reports
Researchers detail VirusCascade, an attack vector where adversarial evidence injected into one agent propagates across collaborative LLM recommenders.
Inspect the evidence
- Inclusion basis
- Enterprise AI
- Publisher and source type
- arXiv cs.LG — Machine Learning · RESEARCH
- Published by source
- 2 October 2026
- Collected by OneBench
- 3 Oct 2026, 03:02 UK
Stored source excerpt
arXiv:2609.38270v1 Announce Type: cross Abstract: Advancing beyond traditional static scoring models, LLM-powered agentic recommender systems (LLM-ARS) instantiate users and items as autonomous agents, whose semantic…
Short excerpt from the collected text, not the full source. Use the source link to read it in context.
The factual summary is a OneBench synthesis, not a quotation or independent verification. Collection time is not publication time. Open the source for its full context; related reporting can share the same underlying announcement.
OneBench interpretation
Institutional assessment
So what
Multi-agent systems using collaborative reflection risk single-point data poisoning that propagates across internal recommender loops.
Do what
Review adversarial robustness controls with the team responsible for agentic recommendation architectures before deploying collaborative workflows.