Before Agents Speak: Pre-hoc Failure Risk Inference in Multi-Agent Systems

2026-07-29

AI Summary

This work addresses the critical challenge of cascading failures in large language model–driven multi-agent systems, where localized hallucinations can rapidly propagate through communication links. Existing approaches predominantly rely on lagging post-hoc detection mechanisms, which are ill-suited for real-time mitigation. To overcome this limitation, the paper introduces HalluProp, a novel proactive hallucination risk reasoning framework tailored for multi-agent settings. HalluProp evaluates intrinsic agent-level hallucination risk via fine-grained semantic alignment, models risk propagation by integrating semantic influence with communication topology, and fuses these signals through a differentiable Noisy-OR mechanism to enable system-wide diagnosis. Experimental results demonstrate that HalluProp achieves an average AUROC of 84.6% across multiple tasks and offers over 65× faster diagnosis than post-hoc methods, enabling sub-second identification of faulty agents with high precision.

Abstract

LLM-based multi-agent systems (MAS) have exhibited remarkable capabilities in collaborative reasoning and decision-making, yet their interconnected communications introduce new systemic risk: localized hallucinations can propagate along agent communication chain, amplify through interactions, and ultimately trigger cascading failures. Existing countermeasures predominantly follow a post-hoc paradigm, identifying failures only after unsafe behaviors emerge, by which time harmful effects may have already spread throughout the agent network. To tackle this problem, we investigate a complementary pre-hoc approach and propose HalluProp, a Propagation-aware Hallucination inference framework that estimates individual agent failures and emergent system-level hallucination risks before inter-agent interaction. First, we model intrinsic hallucination risks by identifying fine-grained semantic misalignment between agent roles and task queries. We then characterize inter-agent risk propagation by modeling both semantic influence and communication topology. Finally, we integrate these two risks via a differentiable Noisy-OR inference mechanism to derive a systemic diagnosis. Extensive experiments show that HalluProp accurately localizes faulty agents, achieving an average AUROC of 84.6%, while enabling sub-second diagnosis with over $65\times$ speedup over post-hoc methods. By facilitating early intervention through upstream screening, HalluProp effectively complements post-hoc methods, highlighting the potential of pre-hoc risk inference for building more reliable multi-agent systems.

Problem

• multi-agent systems
• hallucination propagation
• cascading failures
• pre-hoc risk inference
• systemic risk

Innovation

• pre-hoc risk inference
• hallucination propagation
• multi-agent systems
• semantic misalignment
• Noisy-OR inference