Collective Hallucination in Multi-Agent LLMs:Modeling and Defense

• URL: http://arxiv.org/abs/2606.07941v1
• Date: Sat, 06 Jun 2026 02:04:03 GMT
• Title: Collective Hallucination in Multi-Agent LLMs:Modeling and Defense
• Abstract summary: Hallucinations in large language models (LLMs) create heightened risks in multi-agent settings.<n>This paper models hallucination as a system-level, time-evolving process across a network of interacting LLM agents.<n>We introduce an interaction-aware control method that combines confidence-weighted aggregation, adaptive impact regulation, external claim verification, and selective isolation of unreliable agents.
• Score: 0.5076419064097734
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Hallucinations in large language models (LLMs) create heightened risks in multi-agent settings, where recursive agent interactions can propagate, reinforce, and amplify unsupported claims. This paper models hallucination as a system-level, time-evolving process across a network of interacting LLM agents, where nodes represent agents and edges encode information exchange. The proposed formulation captures how hallucinated claims diffuse through communication topologies, intensify under adversarial perturbations, and affect collective reliability across reasoning rounds. To suppress error propagation, we introduce an interaction-aware control method that combines confidence-weighted aggregation, adaptive impact regulation, external claim verification, and selective isolation of unreliable agents. Experiments on TruthfulQA and TriviaQA show that the proposed method reduces hallucination by up to 39.0% relative to undefended multi-agent reasoning, improves factual accuracy from 0.79 to 0.87, and increases semantic consistency from 0.75 to 0.84. Under adversarial conditions, the method limits hallucination amplification to 1.08, compared with 1.45 without adaptive control, maintaining stable collective behavior across recursive interaction rounds. These results indicate that hallucination in multi-agent LLM systems is governed by both individual model reliability and system-level interaction dynamics, including communication topology, confidence coupling, and recursive information flow.

Related papers

• Before Agents Speak: Pre-hoc Failure Risk Inference in Multi-Agent Systems [16.529918071209643]
  LLM-based multi-agent systems (MAS) have exhibited remarkable capabilities in collaborative reasoning and decision-making.<n> localized hallucinations can propagate along agent communication chain, amplify through interactions, and ultimately trigger cascading failures.<n>Existing countermeasures predominantly follow a post-hoc paradigm, identifying failures only after unsafe behaviors emerge.<n>HalluProp estimates individual agent failures and emergent system-level hallucination risks before inter-agent interaction.
  arXiv  Detail & Related papers  (2026-07-29T12:26:41Z)
• Swarm-Driven Multi-Agent Reasoning for Smart City Security [5.880950534553082]
  This work presents TPSC-Sec, an LLM-based multi-agent approach for stable security reasoning in smart cities.<n>TPSC-Sec decomposes analysis across specialized agents that inspect traffic behavior, protocol interactions, identity usage, and temporal attack progression.<n>Experiments over 500 runs show that TPSC-Sec achieves a high consensus acceptance rate of 0.97 plus or minus 0.02.
  arXiv  Detail & Related papers  (2026-07-03T23:04:48Z)
• Hallucination as Context Drift: Synchronization Protocols for Multi-Agent LLM Systems [0.0]
  Multi-agent LLM systems produce hallucinated outputs that cannot be explained by model deficiencies alone.<n>When agents enter a collaborative task with mismatched or stale representations of shared world state, their joint reasoning produces contradictions that manifest as hallucination.<n>We propose the Shared State Verification Protocol (SSVP), which lets agents periodically exchange compressed state summaries and flag high-divergence conditions before joint reasoning.
  arXiv  Detail & Related papers  (2026-06-19T18:17:45Z)
• Hallucination Cascade: Analyzing Error Propagation in Multi-Agent LLM Systems [6.047611029478951]
  We conduct 500 cascade experiments across 10 knowledge domains using GPT-5.3, DeepSeek-V3, and LLaMA-3-70B-Instruct.<n>We show that deeper cascades reduce the normalized hallucination score from 0.422 at the first agent to 0.272 at the final agent in 3-agent chains.<n>This reduction is accompanied by a decline in factual accuracy from 0.789 to 0.769, revealing a trade-off between hallucination suppression and factual preservation.
  arXiv  Detail & Related papers  (2026-06-06T01:56:55Z)
• Robust Multi-Agent LLMs under Byzantine Faults [12.778788735297288]
  Large language model (LLM) agents increasingly collaborate over peer-to-peer networks to improve their reliability.<n> unreliable or Byzantine agents may sway neighboring agents toward incorrect conclusions and degrade overall system performance.<n>Existing methods rely on leader-based coordination or self-reported confidence, both of which are susceptible to adversarial manipulation.
  arXiv  Detail & Related papers  (2026-05-09T17:37:43Z)
• Taming Actor-Observer Asymmetry in Agents via Dialectical Alignment [59.536125286960186]
  Multi-agent frameworks assigning specialized roles are increasingly adopted to enable self-reflection and mutual auditing.<n>We find it simultaneously induces a human-like cognitive bias known as Actor-Observer Asymmetry.<n>ReTAS guides agents to synthesize conflicting viewpoints into an objective consensus.
  arXiv  Detail & Related papers  (2026-04-21T15:05:58Z)
• AffectAgent: Collaborative Multi-Agent Reasoning for Retrieval-Augmented Multimodal Emotion Recognition [62.16431420189863]
  LLM-based multimodal emotion recognition relies on static parametric memory and often hallucinates when interpreting nuanced affective states.<n>We introduce AffectAgent, an affect-oriented multi-agent retrieval-augmented generation framework.<n>AffectAgent comprises three jointly optimized agents, namely a query planner, an evidence filter, and an emotion generator.
  arXiv  Detail & Related papers  (2026-04-14T13:49:19Z)
• How Independent are Large Language Models? A Statistical Framework for Auditing Behavioral Entanglement and Reweighting Verifier Ensembles [46.63622714488747]
  Shared pretraining data, distillation, and alignment pipelines can induce hidden behavioral dependencies, latent entanglement.<n>In practice, this manifests as correlated reasoning patterns and synchronized failures.<n>We develop a statistical framework for auditing behavioral entanglement among black-box language models.
  arXiv  Detail & Related papers  (2026-04-08T23:32:06Z)
• Information Fidelity in Tool-Using LLM Agents: A Martingale Analysis of the Model Context Protocol [69.11739400975445]
  We introduce the first theoretical framework for analyzing error accumulation in Model Context Protocol (MCP) agents.<n>We show that cumulative distortion exhibits linear growth and high-probability deviations bounded by $O(sqrtT)$.<n>Key findings include: semantic weighting reduces distortion by 80%, and periodic re-grounding approximately every 9 steps suffices for error control.
  arXiv  Detail & Related papers  (2026-02-10T21:08:53Z)
• InEx: Hallucination Mitigation via Introspection and Cross-Modal Multi-Agent Collaboration [6.103123418191468]
  InEx is a training-free, multi-agent framework designed to autonomously mitigate hallucination.<n>InEx consistently outperforms existing methods, achieving 4%-27% gains on general and hallucination benchmarks.
  arXiv  Detail & Related papers  (2025-12-02T17:59:52Z)
• DeLeaker: Dynamic Inference-Time Reweighting For Semantic Leakage Mitigation in Text-to-Image Models [55.30555646945055]
  Text-to-Image (T2I) models are vulnerable to semantic leakage.<n>We introduce DeLeaker, a lightweight approach that mitigates leakage by directly intervening on the model's attention maps.<n>SLIM is the first dataset dedicated to semantic leakage.
  arXiv  Detail & Related papers  (2025-10-16T17:39:21Z)
• Let's Measure Information Step-by-Step: LLM-Based Evaluation Beyond Vibes [14.371259136517802]
  We study robustness of AI systems without ground truth by exploiting a link between strategic gaming and information loss.<n>We analyze which information-theoretic mechanisms resist adversarial bounds, extending finite-sample manipulation to show that bounded f-divergences maintain under attacks.
  arXiv  Detail & Related papers  (2025-08-07T15:11:43Z)
• MIRAGE-Bench: LLM Agent is Hallucinating and Where to Find Them [52.764019220214344]
  Hallucinations pose critical risks for large language model (LLM)-based agents.<n>We present MIRAGE-Bench, the first unified benchmark for eliciting and evaluating hallucinations in interactive environments.
  arXiv  Detail & Related papers  (2025-07-28T17:38:29Z)
• Collaborative Value Function Estimation Under Model Mismatch: A Federated Temporal Difference Analysis [55.13545823385091]
  Federated reinforcement learning (FedRL) enables collaborative learning while preserving data privacy by preventing direct data exchange between agents.<n>In real-world applications, each agent may experience slightly different transition dynamics, leading to inherent model mismatches.<n>We show that even moderate levels of information sharing significantly mitigate environment-specific errors.
  arXiv  Detail & Related papers  (2025-03-21T18:06:28Z)

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.