MAGPIE: A benchmark for Multi-AGent contextual PrIvacy Evaluation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Juneja, Gurusha, Pasupulati, Jayanth Naga Sai, Albalak, Alon, Hua, Wenyue, Wang, William Yang
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908598849765376
author Juneja, Gurusha
Pasupulati, Jayanth Naga Sai
Albalak, Alon
Hua, Wenyue
Wang, William Yang
author_facet Juneja, Gurusha
Pasupulati, Jayanth Naga Sai
Albalak, Alon
Hua, Wenyue
Wang, William Yang
contents A core challenge for autonomous LLM agents in collaborative settings is balancing robust privacy understanding and preservation alongside task efficacy. Existing privacy benchmarks only focus on simplistic, single-turn interactions where private information can be trivially omitted without affecting task outcomes. In this paper, we introduce MAGPIE (Multi-AGent contextual PrIvacy Evaluation), a novel benchmark of 200 high-stakes tasks designed to evaluate privacy understanding and preservation in multi-agent collaborative, non-adversarial scenarios. MAGPIE integrates private information as essential for task resolution, forcing agents to balance effective collaboration with strategic information control. Our evaluation reveals that state-of-the-art agents, including GPT-5 and Gemini 2.5-Pro, exhibit significant privacy leakage, with Gemini 2.5-Pro leaking up to 50.7% and GPT-5 up to 35.1% of the sensitive information even when explicitly instructed not to. Moreover, these agents struggle to achieve consensus or task completion and often resort to undesirable behaviors such as manipulation and power-seeking (e.g., Gemini 2.5-Pro demonstrating manipulation in 38.2% of the cases). These findings underscore that current LLM agents lack robust privacy understanding and are not yet adequately aligned to simultaneously preserve privacy and maintain effective collaboration in complex environments.
format Preprint
id arxiv_https___arxiv_org_abs_2510_15186
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle MAGPIE: A benchmark for Multi-AGent contextual PrIvacy Evaluation
Juneja, Gurusha
Pasupulati, Jayanth Naga Sai
Albalak, Alon
Hua, Wenyue
Wang, William Yang
Cryptography and Security
Computation and Language
A core challenge for autonomous LLM agents in collaborative settings is balancing robust privacy understanding and preservation alongside task efficacy. Existing privacy benchmarks only focus on simplistic, single-turn interactions where private information can be trivially omitted without affecting task outcomes. In this paper, we introduce MAGPIE (Multi-AGent contextual PrIvacy Evaluation), a novel benchmark of 200 high-stakes tasks designed to evaluate privacy understanding and preservation in multi-agent collaborative, non-adversarial scenarios. MAGPIE integrates private information as essential for task resolution, forcing agents to balance effective collaboration with strategic information control. Our evaluation reveals that state-of-the-art agents, including GPT-5 and Gemini 2.5-Pro, exhibit significant privacy leakage, with Gemini 2.5-Pro leaking up to 50.7% and GPT-5 up to 35.1% of the sensitive information even when explicitly instructed not to. Moreover, these agents struggle to achieve consensus or task completion and often resort to undesirable behaviors such as manipulation and power-seeking (e.g., Gemini 2.5-Pro demonstrating manipulation in 38.2% of the cases). These findings underscore that current LLM agents lack robust privacy understanding and are not yet adequately aligned to simultaneously preserve privacy and maintain effective collaboration in complex environments.
title MAGPIE: A benchmark for Multi-AGent contextual PrIvacy Evaluation
topic Cryptography and Security
Computation and Language
url https://arxiv.org/abs/2510.15186