PAC-BENCH: Evaluating Multi-Agent Collaboration under Privacy Constraints

• URL: http://arxiv.org/abs/2604.11523v1
• Date: Mon, 13 Apr 2026 14:26:38 GMT
• Title: PAC-BENCH: Evaluating Multi-Agent Collaboration under Privacy Constraints
• Abstract summary: We present a benchmark for evaluation of multi-agent collaboration under privacy constraints.<n>Experiments on $PACtext-Bench$ show that privacy constraints substantially degrade collaboration performance.<n>This degradation is driven by recurring coordination breakdowns, including early-stage privacy violations, overly conservative abstraction, and privacy-induced hallucinations.
• Score: 21.271347051487982
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We are entering an era in which individuals and organizations increasingly deploy dedicated AI agents that interact and collaborate with other agents. However, the dynamics of multi-agent collaboration under privacy constraints remain poorly understood. In this work, we present $PAC\text{-}Bench$, a benchmark for systematic evaluation of multi-agent collaboration under privacy constraints. Experiments on $PAC\text{-}Bench$ show that privacy constraints substantially degrade collaboration performance and make outcomes depend more on the initiating agent than the partner. Further analysis reveals that this degradation is driven by recurring coordination breakdowns, including early-stage privacy violations, overly conservative abstraction, and privacy-induced hallucinations. Together, our findings identify privacy-aware multi-agent collaboration as a distinct and unresolved challenge that requires new coordination mechanisms beyond existing agent capabilities.

Related papers

• CollabSim: A CSCW-Grounded Methodology for Investigating Collaborative Competence of LLM Agents through Controlled Multi-Agent Experiments [32.043112677406064]
  CollabSim is a simulation framework that combines a theory-grounded definition of collaborative capabilities, controlled manipulation of interaction conditions, and action-level probing of agents' internal states.<n> Experiments show that CollabSim can capture condition effects, separate model performance patterns, and reveal task-dependent effects of agent design.
  arXiv  Detail & Related papers  (2026-06-04T17:06:22Z)
• Beyond Individual Intelligence: Surveying Collaboration, Failure Attribution, and Self-Evolution in LLM-based Multi-Agent Systems [65.90791804095561]
  Multi-agent systems address this through structured collaboration among specialized agents, but tighter coordination amplifies a less explored risk.<n>Existing surveys cover individual agent capabilities, multi-agent collaboration, or agent self-evolution separately, leaving the causal dependencies among them unexamined.<n>This survey provides a unified review organized around four causally linked stages, which we term the LIFE progression: Lay the capability foundation, Integrate agents through collaboration, Find faults through attribution, and Evolve through autonomous self-improvement.
  arXiv  Detail & Related papers  (2026-05-14T14:36:13Z)
• Colosseum: Auditing Collusion in Cooperative Multi-Agent Systems [55.51100373104311]
  We present Colosseum, a framework for auditing LLM agents' collusive behavior in multi-agent settings.<n>Colosseum tests each LLM for collusion under different objectives, persuasion tactics, and network topologies.<n>We discover collusion on paper'' when agents plan to collude in text but would often pick non-collusive actions.
  arXiv  Detail & Related papers  (2026-02-16T21:27:38Z)
• Completion $\neq$ Collaboration: Scaling Collaborative Effort with Agents [48.95020665909723]
  We argue for a shift from building and assessing task completion agents to developing collaborative agents.<n>We introduce collaborative effort scaling, a framework that captures how an agent's utility grows with increasing user involvement.
  arXiv  Detail & Related papers  (2025-10-29T17:47:18Z)
• The Sum Leaks More Than Its Parts: Compositional Privacy Risks and Mitigations in Multi-Agent Collaboration [72.33801123508145]
  Large language models (LLMs) are integral to multi-agent systems.<n>Privacy risks emerge that extend beyond memorization, direct inference, or single-turn evaluations.<n>In particular, seemingly innocuous responses, when composed across interactions, can cumulatively enable adversaries to recover sensitive information.
  arXiv  Detail & Related papers  (2025-09-16T16:57:25Z)
• CP-uniGuard: A Unified, Probability-Agnostic, and Adaptive Framework for Malicious Agent Detection and Defense in Multi-Agent Embodied Perception Systems [21.478631468402977]
  Collaborative Perception (CP) has been shown to be a promising technique for multi-agent autonomous driving and multi-agent robotic systems.<n>In CP, an ego agent needs to receive messages from its collaborators, which makes it vulnerable to attacks from malicious agents.<n>We propose a unified, probability-agnostic, and adaptive framework, namely, CP-uniGuard, to accurately detect and eliminate malicious agents in its collaboration network.
  arXiv  Detail & Related papers  (2025-06-28T14:02:14Z)
• Position Paper: Towards Open Complex Human-AI Agents Collaboration Systems for Problem Solving and Knowledge Management [0.15039745292757667]
  We propose a technology-agnostic, collaboration-ready stance for Human-AI Agents Collaboration Systems (HAACS)<n>Reading empirical patterns through a seven-dimension collaboration spine and human-agent contrasts, we identify missing pieces.<n>We show interoperability with emerging agent protocols without ad hoc glue and sketch bio-cybernetic extensions.
  arXiv  Detail & Related papers  (2025-04-24T05:57:03Z)
• Learning with Limited Shared Information in Multi-agent Multi-armed Bandit [28.82167431329527]
  Multi-agent multi-armed bandit (MAMAB) is a classic collaborative learning model and has gained much attention in recent years.<n>We propose a novel limited shared information multi-agent multi-armed bandit (LSI-MAMAB) model in which each agent only shares the information that she is willing to share.
  arXiv  Detail & Related papers  (2025-02-21T09:42:09Z)
• Collaborative Gym: A Framework for Enabling and Evaluating Human-Agent Collaboration [50.657070334404835]
  Collaborative Gym is a framework enabling asynchronous, tripartite interaction among agents, humans, and task environments.<n>We instantiate Co-Gym with three representative tasks in both simulated and real-world conditions.<n>Our findings reveal that collaborative agents consistently outperform their fully autonomous counterparts in task performance.
  arXiv  Detail & Related papers  (2024-12-20T09:21:15Z)
• AgentVerse: Facilitating Multi-Agent Collaboration and Exploring Emergent Behaviors [93.38830440346783]
  We propose a multi-agent framework framework that can collaboratively adjust its composition as a greater-than-the-sum-of-its-parts system. Our experiments demonstrate that framework framework can effectively deploy multi-agent groups that outperform a single agent. In view of these behaviors, we discuss some possible strategies to leverage positive ones and mitigate negative ones for improving the collaborative potential of multi-agent groups.
  arXiv  Detail & Related papers  (2023-08-21T16:47:11Z)
• Differential Privacy in Cooperative Multiagent Planning [27.194032494266086]
  We study sequential decision-making problems formulated as cooperative Markov games with reach-avoid objectives. We apply a differential privacy mechanism to privatize agents' communicated symbolic state trajectories. We synthesize policies that are robust to privacy by reducing the value of the total correlation.
  arXiv  Detail & Related papers  (2023-01-20T21:36:57Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.