Aetheria: A multimodal interpretable content safety framework based on multi-agent debate and collaboration

2025/12/02 by Yuxiang He, Jian Zhao, He, Yuxiang +17
Computer Science · Social Sciences · #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #Hate Speech and Cyberbullying Detection #Misinformation and Its Impacts #Topic Modeling

paper · pdf · doi:10.48550/arxiv.2512.02530

openalex publication_date 2025/12/02 · openalex created_date 2025/12/04 · openalex updated_date 2026/07/28

Abstract

The exponential growth of digital content presents significant challenges for content safety. Current moderation systems, often based on single models or fixed pipelines, exhibit limitations in identifying implicit risks and providing interpretable judgment processes. To address these issues, we propose Aetheria, a multimodal interpretable content safety framework based on multi-agent debate and collaboration.Employing a collaborative architecture of five core agents, Aetheria conducts in-depth analysis and adjudication of multimodal content through a dynamic, mutually persuasive debate mechanism, which is grounded by RAG-based knowledge retrieval.Comprehensive experiments on our proposed benchmark (AIR-Bench) validate that Aetheria not only generates detailed and traceable audit reports but also demonstrates significant advantages over baselines in overall content safety accuracy, especially in the identification of implicit risks. This framework establishes a transparent and interpretable paradigm, significantly advancing the field of trustworthy AI content moderation.

Citations

• TeleAI-Safety: A comprehensive LLM jailbreaking benchmark towards attacks, defenses, and evaluations
• Efficient LLM Safety Evaluation through Multi-Agent Debate
• OrthAlign: Orthogonal Subspace Decomposition for Non-Interfering Multi-Objective Alignment
• RADAR: A Risk-Aware Dynamic Multi-Agent Framework for LLM Safety Evaluation via Role-Specialized Collaboration
• When Safe Unimodal Inputs Collide: Optimizing Reasoning Chains for Cross-Modal Safety in Multimodal Large Language Models
• Never compromise with vulnerabilities: a comprehensive survey on AI governance
• USB: A Comprehensive and Unified Safety Evaluation Benchmark for Multimodal Large Language Models
• WildGuard: Open One-Stop Moderation Tools for Safety Risks, Jailbreaks, and Refusals of LLMs
• AI Supported Degradation of the Self Concept: A Theoretical Framework Grounded in Established Cognitive and Computational Mechanisms
• Trustworthy LLMs: a Survey and Guideline for Evaluating Large Language Models' Alignment
• Simple synthetic data reduces sycophancy in large language models
• MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework
• Encouraging Divergent Thinking in Large Language Models through Multi-Agent Debate
• Improving Factuality and Reasoning in Language Models through Multiagent Debate
• Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Prompting
• ToxiGen: A Large-Scale Machine-Generated Dataset for Adversarial and\n Implicit Hate Speech Detection
• On the Opportunities and Risks of Foundation Models
• On the Dangers of Stochastic Parrots

Cited by

• Never compromise with vulnerabilities: a comprehensive survey on AI governance

Related

• AEGIS: Online Adaptive AI Content Safety Moderation with Ensemble of LLM Experts
• AI vs. Human Moderators: A Comparative Evaluation of Multimodal LLMs in Content Moderation for Brand Safety
• A Different Approach to AI Safety: Proceedings from the Columbia Convening on Openness in Artificial Intelligence and AI Safety
• Class-RAG: Real-Time Content Moderation with Retrieval Augmented Generation
• RedDebate: Safer Responses Through Multi-Agent Red Teaming Debates