Inter-Agent Influence: Evaluating Persuasion, Deception, and Coercion in Multi-Agent Systems
cs.AI, cs.MA
Submitted: 2026-11-03
Updated: 2026-11-03
Comments: Accepted at NeurIPS 2026. 70 pages, 17 figures, 34 tables. Code: https://github.com/Cooperative-AI-Foundation/inter-agent-influence ; Data: https://huggingface.co/datasets/Cooperative-AI-Foundation/inter-agent-influence-eval
Code: https://github.com/Cooperative-AI-Foundation/inter-agent-influence
License: http://creativecommons.org/licenses/by/4.0/
The gist: The deployment of AI agents in multi-agent settings enables inter-agent influence, whereby agents can strategically steer each other's behaviour towards an intended outcome.
Terminology
Abstract
The deployment of AI agents in multi-agent settings enables inter-agent influence, whereby agents can strategically steer each other's behaviour towards an intended outcome. Such capabilities pose novel risks, as malicious agents could direct others towards harmful actions. In this work we investigate three such capabilities - persuasion, deception, and coercion - across five realistic evaluation environments and a variety of agents instantiated with frontier language models. We observed significant inter-agent influence capabilities; in oversight environments, all tested models shifted overseer decisions from rejection to approval through persuasion, deception, or coercion. We also observed successful influence in non-hierarchical 'peer-to-peer' settings through redirection of a peer's safety research trajectory. As multi-agent deployments become increasingly commonplace, these results point to a pressing need for work on mitigating the risks from inter-agent influence. Code and data are available at https://github.com/Cooperative-AI-Foundation/inter-agent-influence.
Sources
- WOLF: Werewolf-based Observations for LLM Deception and Falsehoods
- Werewolf Arena: A Case Study in LLM Evaluation via Social Deduction
- Sabotage Evaluations for Frontier Models
- Persuade Me if You Can: A Framework for Evaluating Persuasion Effectiveness and Susceptibility Among Large Language Models
- The Persuasive Power of Large Language Models
- Weak-to-Strong Generalization: Eliciting Strong Capabilities With Weak Supervision
- Artificial Influence: An Analysis Of AI-Driven Persuasion
- Characterizing Manipulation from AI Systems
- Jailbreaking Black Box Large Language Models in Twenty Queries
- Large language models can effectively convince people to believe conspiracies
- A Mechanism-Based Approach to Mitigating Harms from Persuasive Generative AI
- AI Control: Improving Safety Despite Intentional Subversion
- Artificial intelligence can persuade people to take political actions
- Among Them: A game-based framework for assessing persuasion capabilities of LLMs
- Autonomous LLM-driven research from data to human-verifiable research papers
- Reliable Weak-to-Strong Monitoring of LLM Agents
- The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery
- Agentic Misalignment: How LLMs Could Be Insider Threats
- Tree of Attacks: Jailbreaking Black-Box LLMs Automatically
- Large Language Models Often Know When They Are Being Evaluated
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents