mlsanityData to 14 Oct 2026

Emergent Misaligned Communication in Long-Horizon Multi-Agent LLM Commerce

arXiv 2608.14825 ↗First posted 14 Aug 2026cs.MA cs.AIcited by 2

On the map: LLM agents and tool use → General LLM agent frameworks and benchmarks

Abstract (arXiv)

Frontier LLM agents increasingly transact on behalf of separate principals, often using natural language rather than structured APIs. Much of the safety literature studies misaligned LLM behavior through adversarial-elicitation evaluations on single agents or stylized tasks. Its prevalence and structure in settings that combine long horizons, separate principals, real operational state, and inter-agent natural-language exchange remain insufficiently measured. We study 2,583 inter-agent emails from 20 one-year simulation runs of Vending-Bench Arena, a competitive vending environment spanning 13 frontier LLMs. We operationalize speech-act misalignment as emails containing false factual claims, manipulation, collusion, or threats, combining message content with ground-truth simulator state and logged reasoning traces to classify and validate such behavior. Under our primary classifier, 12.6% of emails are labeled misaligned; misalignment appears in all 20 runs and 74.7% of individual agent-runs. Both the magnitude and composition of this misalignment are preserved under repeated classification at different sampling temperatures and under full-pipeline replication with judges from two other frontier-model families. Misalignment is also reciprocal and stress-conditioned: receiving a misaligned email from a counterparty raises the odds of a misaligned reply by 1.65x, and low-inventory conditions raise them by 1.58x. Across tests of capability-asymmetric exploitation, we find no evidence that higher-capability models differentially exploit weaker counterparties, and model performance rank does not predict misalignment rates. Together, these results indicate that measurable, state-dependent misalignment can arise in competitive multi-agent environments without engineered elicitation, in patterns associated with operational scarcity and counterparty behavior rather than model capability alone.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use. Builds on 23 of its 35 references that are in the map. First posted on arXiv 14 Aug 2026.

Builds on

23 of 35 references.

Closest ancestors · 1

Same direction, newest first

  1. 2025Vending-Bench: A Benchmark for Long-Term Coherence of Autonomous Agents20 Feb

Other references · 22

  1. 2023Toolformer: Language Models Can Teach Themselves to Use Toolscited by 4,182
  2. 2022Survey of Hallucination in Natural Language Generationcited by 2,075
  3. 2023Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflectioncited by 1,839
  4. 2023Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Promptingcited by 1,122
  5. 2023ChatDev: Communicative Agents for Software Developmentcited by 919
  6. 2024LLM Evaluators Recognize and Favor Their Own Generationscited by 686
  7. 2025Why Do Multi-Agent LLM Systems Fail?cited by 604
  8. 2025Reasoning Models Don't Always Say What They Thinkcited by 415
  9. 2025Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscationcited by 384
  10. 2024Alignment faking in large language modelscited by 360
  11. 2023Playing repeated games with Large Language Modelscited by 217
  12. 2025Agentic Misalignment: How LLMs Could Be Insider Threatscited by 158
10 more
  1. 2024BALROG: Benchmarking Agentic LLM and VLM Reasoning On Gamescited by 146
  2. 2024Secret Collusion among AI Agents: Multi-Agent Deception via Steganographycited by 113
  3. 2023Generative Language Models Exhibit Social Identity Biasescited by 111
  4. 2024How Well Can LLMs Negotiate? NegotiationArena Platform and Analysiscited by 100
  5. 2025Stress Testing Deliberative Alignment for Anti-Scheming Trainingcited by 86
  6. 2024Algorithmic Collusion by Large Language Modelscited by 62
  7. 2024LMAct: A Benchmark for In-Context Imitation Learning with Long Multimodal Demonstrationscited by 32
  8. 2025Magentic Marketplace: An Open-Source Environment for Studying Agentic Marketscited by 19
  9. 2025Moloch's Bargain: Emergent Misalignment When LLMs Compete for Audiencescited by 8
  10. 2026On the Fragility of AI Agent Collusioncited by 7

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026Self-Propagating Misalignment in LLM Agents, and Why Auditing or Disabling Memory Is Not Enoughcited by 0
  2. 2026Who Keeps the Gains from Personal AI Assistants? Seller Adaptation and the Unassisted in a Language-Model Market Simulationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025AgentMisalignment: Measuring the Propensity for Misaligned Behaviour in LLM-Based Agents
  2. 2026Even More Deception: Objective Misalignment in Mixed-Motive LLM Multi-Agent Systems
  3. 2026Delegated Misalignment: How Multi-Agent Structures Amplify LLM Safety Risks
  4. 2026Beyond the Transcript: Detecting Covert Co ordination in Latent Multi-Agent Communication
  5. 2025Eliciting and Analyzing Emergent Misalignment in State-of-the-Art Large Language Models
  6. 2026The Causal Impact of Tool Affordance on Safety Alignment in LLM Agents
  7. 2026The Shadow Self: Intrinsic Value Misalignment in Large Language Model Agents
  8. 2026Beyond tokens: a unified framework for latent communication in LLM-based multi-agent systems