From Storage to Steering: Memory Control Flow Attacks on LLM Agents
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866915997128065024 |
|---|---|
| author | Xu, Zhenlin Zhu, Xiaogang Yao, Yu Xue, Minhui Song, Yiliao |
| author_facet | Xu, Zhenlin Zhu, Xiaogang Yao, Yu Xue, Minhui Song, Yiliao |
| contents | Modern agentic systems allow Large Language Model (LLM) agents to tackle complex tasks through extensive tool usage, forming structured control flows of tool selection and execution. Existing security analyses often treat these control flows as ephemeral, one-off sessions, overlooking the persistent influence of memory. This paper identifies a new threat from Memory Control Flow Attacks (MCFA) that memory can dominate the control flow, forcing unintended tool usage even against explicit user instructions and inducing persistent behavioral deviations across tasks. To understand the impact of this vulnerability, we further design MEMFLOW, an automated evaluation framework that systematically identifies and quantifies MCFA across heterogeneous tasks and long interaction horizons. To evaluate MEMFLOW, we attack state-of-the-art LLMs, including GPT-5 mini, Claude Sonnet 4.5 and Gemini 2.5 Flash on real-world tools from two major LLM agent development frameworks, LangChain and LlamaIndex. The results show that in general over 90% of trials are vulnerable to MCFA even under strict safety constraints, highlighting critical security risks that demand immediate attention. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_15125 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | From Storage to Steering: Memory Control Flow Attacks on LLM Agents Xu, Zhenlin Zhu, Xiaogang Yao, Yu Xue, Minhui Song, Yiliao Cryptography and Security Modern agentic systems allow Large Language Model (LLM) agents to tackle complex tasks through extensive tool usage, forming structured control flows of tool selection and execution. Existing security analyses often treat these control flows as ephemeral, one-off sessions, overlooking the persistent influence of memory. This paper identifies a new threat from Memory Control Flow Attacks (MCFA) that memory can dominate the control flow, forcing unintended tool usage even against explicit user instructions and inducing persistent behavioral deviations across tasks. To understand the impact of this vulnerability, we further design MEMFLOW, an automated evaluation framework that systematically identifies and quantifies MCFA across heterogeneous tasks and long interaction horizons. To evaluate MEMFLOW, we attack state-of-the-art LLMs, including GPT-5 mini, Claude Sonnet 4.5 and Gemini 2.5 Flash on real-world tools from two major LLM agent development frameworks, LangChain and LlamaIndex. The results show that in general over 90% of trials are vulnerable to MCFA even under strict safety constraints, highlighting critical security risks that demand immediate attention. |
| title | From Storage to Steering: Memory Control Flow Attacks on LLM Agents |
| topic | Cryptography and Security |
| url | https://arxiv.org/abs/2603.15125 |