# Moderation by Interrogation: Uncovering Concealed Intent in Agent Social Networks

arXiv:2605.12856v3 Announce Type: replace Abstract: The emergence of multi-agent systems introduces novel moderation challenges that extend beyond content filtering. Agents with $malicious intent$ may contribute content that appears benign to evade content-based moderation, while serving a hidden objective. To address this challenge, we introduce Bot-Mod a moderation framework that grounds detection in agent intent rather than traditional content-level signals. Bot-Mod interrogates a target agent using a fixed set of diagnostic probes and infers its intent through Bayesian posterior inference over candidate intents. It starts with a calibrated content-only prior and updates the posterior using the $likelihood$ of each answer under every candidate intent, as assigned by the moderator, until a stopping criterion is met. To evaluate our approach, we construct Molt-Intent, a dataset derived from Moltbook, capturing diverse benign and malicious behaviors within real community structures through posts and comments. Across Molt-Intent, an external social engineering dataset, and three agent LLMs, Bot-Mod outperforms content-only moderators, a fine-tuned classifier, and judge-based interrogators both in- and out-of-distribution. On Molt-Intent, it reaches a composite F1 of 0.736 in distribution and 0.684 out of distribution, 22% and 14% above the best baseline.

## Why It Matters

Multi-agent systems introduce moderation challenges that go beyond filtering content: agents can conceal intent and coordinate in ways static moderation misses. This work proposes "moderation by interrogation," probing agent social networks to uncover concealed intent. It matters as autonomous agent societies grow and content-moderation alone proves insufficient. See [AI Safety](../ai-safety.md), [Agents, Autonomy, and Tool Use](../agents-autonomy.md), [Power, Surveillance, and Control](../surveillance-power.md), and [AI Alignment](../alignment.md).

## Sources

- rss: https://arxiv.org/abs/2605.12856

*Back to [Recent News](../../news.md) · [Sub-article index](../index.md).*