Hijacking Agent Memory: Stealthy Trojan Attacks Through Conversational Interaction

Abstract

Large language model (LLM) agents increasingly leverage long term memory to support persistent and autonomous task execution. However, this capability also introduces a new attack surface: memory poisoning, where adversaries can inject malicious information to influence future behavior. Existing memory poisoning attacks often assume that injected content can be stored directly in memory, overlooking the selective extraction and rewriting stages in modern memory pipelines. This makes prior method

Categories

Framework mappings

Suggested from the entry's categories.

@misc{wang2026hijacking,
  title = {{Hijacking Agent Memory: Stealthy Trojan Attacks Through Conversational Interaction}},
  author = {Hongtao Wang and Se Yang and Yu Chen and Puzhuo Liu},
  year = {2026},
  month = may,
  eprint = {2605.29960},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2605.29960}
}