---
title: "How Much Can Tool Calls Reveal About You? Privacy Leakage from Tool-Call Traces"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2610.41885v1"
permalink: "https://feed.astralor.com/items/itm_19603a73989e7f788a028c3f"
published: "2026-10-30T15:30:23.000Z"
score: 55
channel: "tech"
tags: ["LLM","AI Agent","推理效率","How Much Can","Tool Calls Reveal","About You","Privacy Leakage","Tool-Call Traces Large"]
---

# How Much Can Tool Calls Reveal About You? Privacy Leakage from Tool-Call Traces

## 速览

Large language model (LLM)-based agents interact with external tools through protocols such as the Model Context Protocol (MCP), producing tool-call traces that may expose behavioral signals about users. In this work, w…

## 关键点

- Large language model (LLM)-based agents interact with external tools through protocols such as the Model Context Protocol (MCP), producing tool-call traces that may expose behavioral signals about users.
- In this work, we systematically study the risk of user-profile inference from MCP tool-call traces, where observers infer user attributes from protocol-valid tool interactions without access to user conversations.
- We introduce PrivacyTrace, a benchmark containing 1,000 synthetic user profiles and 4,000 agent trajectories across four domains, generated with real MCP tools from 25 servers.

## 全文

Large language model (LLM)-based agents interact with external tools through protocols such as the Model Context Protocol (MCP), producing tool-call traces that may expose behavioral signals about users. In this work, we systematically study the risk of user-profile inference from MCP tool-call traces, where observers infer user attributes from protocol-valid tool interactions without access to user conversations. We introduce PrivacyTrace, a benchmark containing 1,000 synthetic user profiles and 4,000 agent trajectories across four domains, generated with real MCP tools from 25 servers. Experiments with four mainstream LLM attackers show that tool-call traces enable a 26.25% attack success rate for profile inference, compared with 17.97% from prior-based guessing. Further analysis reveals that tool arguments and returned results provide the main trace-grounded evidence, while linking traces across sessions further amplifies profiling risk. These findings demonstrate that MCP tool-call traces can act as effective behavioral side channels for user profiling, motivating the development of privacy-aware mechanisms for agent-tool interactions. Code and data are available at https://github.com/AI45Lab/PrivacyTrace.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_19603a73989e7f788a028c3f