# GAMBIT: A Three-Mode Benchmark for Adversarial Robustness in Multi-Agent LLM Collectives arXiv: https://arxiv.org/abs/2605.09027 ยท HTML: https://mlsanity.com/paper/2605.09027/ ## Key facts - Cited by 1 papers in the map (data through 14 Oct 2026). - Placed in Agent threats, guardrails and governance, one of 16 directions in LLM agents and tool use. - Builds on 40 of its 49 references that are in the map. - First posted on arXiv 9 May 2026. ## Placed in - LLM agents and tool use / [Agent threats, guardrails and governance](https://mlsanity.com/agents/agent-threats-guardrails-and-governance/) ## Builds on (most cited first, 30 of 40) - [Self-Consistency Improves Chain of Thought Reasoning in Language Models](https://mlsanity.com/paper/2203.11171/) (2022) - [Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks](https://mlsanity.com/paper/1703.03400/) (2017) - [Optuna: A Next-generation Hyperparameter Optimization Framework](https://mlsanity.com/paper/1907.10902/) (2019) - [Improving Factuality and Reasoning in Language Models through Multiagent Debate](https://mlsanity.com/paper/2305.14325/) (2023) - [HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal](https://mlsanity.com/paper/2402.04249/) (2024) - [Red Teaming Language Models with Language Models](https://mlsanity.com/paper/2202.03286/) (2022) - [Large Language Model based Multi-Agents: A Survey of Progress and Challenges](https://mlsanity.com/paper/2402.01680/) (2024) - [Why Do Multi-Agent LLM Systems Fail?](https://mlsanity.com/paper/2503.13657/) (2025) - [JailbreakBench: An Open Robustness Benchmark for Jailbreaking Large Language Models](https://mlsanity.com/paper/2404.01318/) (2024) - [MetaICL: Learning to Learn In Context](https://mlsanity.com/paper/2110.15943/) (2021) - [Mixture-of-Agents Enhances Large Language Model Capabilities](https://mlsanity.com/paper/2406.04692/) (2024) - [Rapid Learning or Feature Reuse? Towards Understanding the Effectiveness of MAML](https://mlsanity.com/paper/1909.09157/) (2019) - [Rainbow Teaming: Open-Ended Generation of Diverse Adversarial Prompts](https://mlsanity.com/paper/2402.16822/) (2024) - [Meta-learning via Language Model In-context Tuning](https://mlsanity.com/paper/2110.07814/) (2021) - [Towards a Science of Scaling Agent Systems](https://mlsanity.com/paper/2512.08296/) (2025) - [Red-Teaming LLM Multi-Agent Systems via Communication Attacks](https://mlsanity.com/paper/2502.14847/) (2025) - [GTBench: Uncovering the Strategic Reasoning Limitations of LLMs via Game-Theoretic Evaluations](https://mlsanity.com/paper/2402.12348/) (2024) - [On the Resilience of LLM-Based Multi-Agent Collaboration with Faulty Agents](https://mlsanity.com/paper/2408.00989/) (2024) - [Flooding Spread of Manipulated Knowledge in LLM-Based Multi-Agent Communities](https://mlsanity.com/paper/2407.07791/) (2024) - [ChessGPT: Bridging Policy Learning and Language Modeling](https://mlsanity.com/paper/2306.09200/) (2023) - [Talk Isn't Always Cheap: Understanding Failure Modes in Multi-Agent Debate](https://mlsanity.com/paper/2509.05396/) (2025) - [G-Safeguard: A Topology-Guided Security Lens and Treatment on LLM-based Multi-agent Systems](https://mlsanity.com/paper/2502.11127/) (2025) - [Learning to Few-Shot Learn Across Diverse Natural Language Classification Tasks](https://mlsanity.com/paper/1911.03863/) (2019) - [MultiAgent Collaboration Attack: Investigating Adversarial Attacks in Large Language Model Collaborations via Debate](https://mlsanity.com/paper/2406.14711/) (2024) - [Rethinking Mixture-of-Agents: Is Mixing Different Large Language Models Beneficial?](https://mlsanity.com/paper/2502.00674/) (2025) - [Amortized Planning with Large-Scale Transformers: A Case Study on Chess](https://mlsanity.com/paper/2402.04494/) (2024) - [Investigating Meta-Learning Algorithms for Low-Resource Natural Language Understanding Tasks](https://mlsanity.com/paper/1908.10423/) (2019) - [BlindGuard: Safeguarding LLM-based Multi-Agent Systems under Unknown Attacks](https://mlsanity.com/paper/2508.08127/) (2025) - [The Traitors: Deception and Trust in Multi-Agent Language Model Simulations](https://mlsanity.com/paper/2505.12923/) (2025) - [Explore the Reasoning Capability of LLMs in the Chess Testbed](https://mlsanity.com/paper/2411.06655/) (2024) ## Built on by (most cited first, 1 of 1) - [SoK: When Safe Agents Fail Together: The Security of Multi Agent LLM Systems](https://mlsanity.com/paper/2609.00595/) (2026) ## Abstract (arXiv) A collective of AI agents collaborating on a task has the potential to outclass any individual agent for that task. We study the robustness of such collectives against possible imposters, i.e., agents that deliberately try to mislead their peers. Since a single imposter could undo the collective's advantage, we need to detect them. We consider two strategies: (i) incorporate imposter detection into the participating agents, or (ii) use a dedicated imposter detector outside the collective. We investigate this empirically on Gambit, a testbed in which 4 reasoning agents deliberate on chess moves. The setting is small but still challenging for frontier models. Chess allows objective, quantitative assessment (via a state-of-the-art chess engine) of both the gain of using a collective and the damage done by imposters. We find that merely warning the agents of potential imposter presence is not beneficial: it degrades decisions when no imposter is present, provokes reactions ranging from self-accusation to scapegoating, inflates token use, and reveals to the imposter how it was uncovered. We therefore recommend a detector that reads the collective's deliberation but never joins it and only returns a verdict. Such a detector must recalibrate to new attack strategies after very few examples, rather than wait for full retraining. In our benchmark, a 3B language model with a meta-trained classification head achieves that: a single gradient step on 20 labeled examples suffices to adapt to an unseen imposter strategy. At matched zero-shot accuracy, this detector yields 8x the adaptation gain of standard finetuning, at 14x lower training cost. We release the Gambit benchmark, with 37,352 labeled deliberations spanning 240 evolved imposter strategies. Code and data: https://anonymous.4open.science/r/gambit.