mlsanityData to 14 Oct 2026

Among Us: A Sandbox for Measuring and Detecting Agentic Deception

arXiv 2504.04072 ↗First posted 5 Apr 2025cs.AI cs.LGcited by 26

On the map: LLM agents and tool use → Social simulation and multi-agent debate

also in 1 more
Abstract (arXiv)

Prior studies on deception in language-based AI agents typically assess whether the agent produces a false statement about a topic, or makes a binary choice prompted by a goal, rather than allowing open-ended deceptive behavior to emerge in pursuit of a longer-term goal. To fix this, we introduce Among Us, a sandbox social deception game where LLM-agents exhibit long-term, open-ended deception as a consequence of the game objectives. While most benchmarks saturate quickly, Among Us can be expected to last much longer, because it is a multi-player game far from equilibrium. Using the sandbox, we evaluate 18 proprietary and open-weight LLMs and uncover a general trend: models trained with RL are comparatively much better at producing deception than detecting it. We evaluate the effectiveness of methods to detect lying and deception: logistic regression on the activations and sparse autoencoders (SAEs). We find that probes trained on a dataset of "pretend you're a dishonest model:.." generalize extremely well out-of-distribution, consistently obtaining AUROCs over 95% even when evaluated just on the deceptive statement, without the chain of thought. We also find two SAE features that work well at deception detection but are unable to steer the model to lie less. We hope our open-sourced sandbox, game logs, and probes serve to anticipate and mitigate deceptive behavior and capabilities in language-based agents.

Key facts. Cited by 26 papers in the map (data through 14 Oct 2026). Placed in Social simulation and multi-agent debate, one of 16 directions in LLM agents and tool use. Builds on 24 of its 38 references that are in the map. First posted on arXiv 5 Apr 2025.

Builds on

24 of 38 references.

References · 24

  1. 2020Measuring Massive Multitask Language Understandingcited by 7,963
  2. 2021TruthfulQA: Measuring How Models Mimic Human Falsehoodscited by 2,975
  3. 2019HuggingFace's Transformers: State-of-the-art Natural Language Processingcited by 2,218
  4. 2023Sparse Autoencoders Find Highly Interpretable Features in Language Modelscited by 1,420
  5. 2016Understanding intermediate layers using linear classifier probescited by 1,353
  6. 2022Discovering Latent Knowledge in Language Models Without Supervisioncited by 767
  7. 2023The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasetscited by 712
  8. 2023Detecting Pretraining Data from Large Language Modelscited by 387
  9. 2024Alignment faking in large language modelscited by 360
  10. 2023Exploring Large Language Models for Communication Games: An Empirical Study on Werewolfcited by 241
  11. 2023AI Control: Improving Safety Despite Intentional Subversioncited by 229
  12. 2024A Survey on Large Language Model-Based Game Agentscited by 112
12 more
  1. 2025Detecting Strategic Deception Using Linear Probescited by 68
  2. 2022Mastering the Game of No-Press Diplomacy via Human-Regularized Reinforcement Learning and Planningcited by 48
  3. 2023Hoodwinked: Deception and Cooperation in a Text-Based Game for Language Modelscited by 47
  4. 2025Gradual Disempowerment: Systemic Existential Risks from Incremental AI Developmentcited by 46
  5. 2024Richelieu: Self-Evolving LLM-Based Agents for AI Diplomacycited by 36
  6. 2024AMONGAGENTS: Evaluating Large Language Models in the Interactive Text-Based Social Deduction Gamecited by 28
  7. 2025Deception in LLMs: Self-Preservation and Autonomous Goals in Large Language Modelscited by 16
  8. 2025Training Language Models for Social Deduction with Multi-Agent Reinforcement Learningcited by 16
  9. 2024Lies, Damned Lies, and Distributional Language Statistics: Persuasion and Deception with Large Language Modelscited by 15
  10. 2025Line Goes Up? Inherent Limitations of Benchmarks for Evaluating Large Language Modelscited by 14
  11. 2024Multi-Agent Training for Pommerman: Curriculum Learning and Population-based Self-Play Approachcited by 3
  12. 2024LLMs are Superior Feedback Providers: Bootstrapping Reasoning for Lie Detection with Self-Generated Feedbackcited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 7; 2026: 19.

26 papers.

Sort:

By area:

  1. 2025TRiSM for Agentic AI: A Review of Trust, Risk, and Security Management in LLM-based Agentic Multi-Agent Systemscited by 59
  2. 2026Lying with Truths: Open-Channel Multi-Agent Collusion for Belief Manipulation via Generative Montagecited by 19
  3. 2025Difficulties with Evaluating a Deception Detector for AIscited by 13
  4. 2025Deceive, Detect, and Disclose: Large Language Models Play Mini-Mafiacited by 9
  5. 2026Among Us: Measuring and Mitigating Malicious Contributions in Model Collaboration Systemscited by 7
  6. 2026Detecting Multi-Agent Collusion Through Multi-Agent Interpretabilitycited by 7
  7. 2025A Mathematical Philosophy of Explanations in Mechanistic Interpretability -- The Strange Science Part I.icited by 6
  8. 2026LieCraft: A Multi-Agent Framework for Evaluating Deceptive Capabilities in Language Modelscited by 5
  9. 2025PhenoAssistant: A Conversational Multi-Agent AI System for Automated Plant Phenotypingcited by 4
  10. 2026Activation Steering for Aligned Open-ended Generation without Sacrificing Coherencecited by 4
  11. 2026Deception and Communication in Autonomous Multi-Agent Systems: An Experimental Study with Among Uscited by 3
  12. 2025Shall We Play a Game? Language Models for Open-ended Wargamescited by 2
14 more
  1. 2026MaMa: A Game-Theoretic Approach for Designing Safe Agentic Systemscited by 2
  2. 2026MafiaScope: Non-Invasive, Time-Resolved Belief Probing for LLM Agents in Social Deduction Gamescited by 2
  3. 2026Coercion and Deception in AI-to-AI Management: An Agentic Benchmark of Unprompted Escalationcited by 2
  4. 2025Debate with Images: Detecting Deceptive Behaviors in Multimodal Large Language Modelscited by 1
  5. 2026ElecTwit: A Framework for Studying Persuasion in Multi-Agent Social Systemscited by 1
  6. 2026Evaluating Large Language Models in a Complex Hidden Role Gamecited by 1
  7. 2026Behind EvoMap: Characterizing a Self-Evolving Agent-to-Agent Collaboration Networkcited by 1
  8. 2026RogueAI: A Reverse Turing Test for Detecting Licensed AI Deception in Dialoguecited by 1
  9. 2026Disentangling Deception and Hallucination Failures in LLMscited by 0
  10. 2026SocialGrid: A Benchmark for Planning and Social Reasoning in Embodied Multi-Agent Systemscited by 0
  11. 2026Even More Deception: Objective Misalignment in Mixed-Motive LLM Multi-Agent Systemscited by 0
  12. 2026Can Agents Deceive? Evaluating Reasoning and Deception in ParliamentBench using a Social Deduction Gamecited by 0
  13. 2026MARBO: Relational Belief Grounding for LLM Agents in Social Deduction Gamescited by 0
  14. 2026MCAIR: Monitoring and Correcting Agent Unfaithfulness via Internal Representationscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Hidden in Plain Text: Measuring LLM Deception Quality Against Human Baselines Using Social Deduction Games
  2. 2026Lying to Win: Assessing LLM Deception through Human-AI Games and Parallel-World Probing
  3. 2025DeceptionBench: A Comprehensive Benchmark for AI Deception Behaviors in Real-world Scenarios
  4. 2026Lies We Can See: Joint Verbal and Non-Verbal Deception by VLM Agents in Embodied Social Interactions
  5. 2026DecepEval: A Benchmark for Evaluating Deception in LLM Agents
  6. 2025The Traitors: Deception and Trust in Multi-Agent Language Model Simulations
  7. 2025When Thinking LLMs Lie: Unveiling the Strategic Deception in Representations of Reasoning Models
  8. 2025The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systems