# Privacy-Preserving Topology-Guided Safety for LLM-Based Multi-Agent Systems via Federated Graph Learning

arXiv: https://arxiv.org/abs/2609.02967 · HTML: https://mlsanity.com/paper/2609.02967/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Agent threats, guardrails and governance, one of 16 directions in LLM agents and tool use.
- Builds on 23 of its 41 references that are in the map.
- First posted on arXiv 2 Sep 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, 23 of 23)

- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://mlsanity.com/paper/1908.10084/) (2019)
- [Graph Attention Networks](https://mlsanity.com/paper/1710.10903/) (2017)
- [Communication-Efficient Learning of Deep Networks from Decentralized Data](https://mlsanity.com/paper/1602.05629/) (2016)
- [Federated Optimization in Heterogeneous Networks](https://mlsanity.com/paper/1812.06127/) (2018)
- [AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation](https://mlsanity.com/paper/2308.08155/) (2023)
- [Improving Factuality and Reasoning in Language Models through Multiagent Debate](https://mlsanity.com/paper/2305.14325/) (2023)
- [Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations](https://mlsanity.com/paper/2312.06674/) (2023)
- [Agent-SafetyBench: Evaluating the Safety of LLM Agents](https://mlsanity.com/paper/2412.14470/) (2024)
- [A Dynamic LLM-Powered Agent Network for Task-Oriented Agent Collaboration](https://mlsanity.com/paper/2310.02170/) (2023)
- [R-Judge: Benchmarking Safety Risk Awareness for LLM Agents](https://mlsanity.com/paper/2401.10019/) (2024)
- [Text Embeddings Reveal (Almost) As Much As Text](https://mlsanity.com/paper/2310.06816/) (2023)
- [FedDC: Federated Learning with Non-IID Data via Local Drift Decoupling and Correction](https://mlsanity.com/paper/2203.11751/) (2022)
- [ShieldAgent: Shielding Agents via Verifiable Safety Policy Reasoning](https://mlsanity.com/paper/2503.22738/) (2025)
- [AgentDojo: A Dynamic Environment to Evaluate Prompt Injection Attacks and Defenses for LLM Agents](https://mlsanity.com/paper/2406.13352/) (2024)
- [Subgraph Federated Learning with Missing Neighbor Generation](https://mlsanity.com/paper/2106.13430/) (2021)
- [GuardAgent: Safeguard LLM Agents by a Guard Agent via Knowledge-Enabled Reasoning](https://mlsanity.com/paper/2406.09187/) (2024)
- [Federated Learning on Non-IID Graphs via Structural Knowledge Sharing](https://mlsanity.com/paper/2211.13009/) (2022)
- [BlindGuard: Safeguarding LLM-based Multi-Agent Systems under Unknown Attacks](https://mlsanity.com/paper/2508.08127/) (2025)
- [GUARDIAN: Safeguarding LLM Multi-Agent Collaborations with Temporal Graph Modeling](https://mlsanity.com/paper/2505.19234/) (2025)
- [Poison-RAG: Adversarial Data Poisoning Attacks on Retrieval-Augmented Generation in Recommender Systems](https://mlsanity.com/paper/2501.11759/) (2025)
- [Federated Graph Semantic and Structural Learning](https://mlsanity.com/paper/2406.18937/) (2024)
- [OpenFGL: A Comprehensive Benchmark for Federated Graph Learning](https://mlsanity.com/paper/2408.16288/) (2024)
- [SafeSieve: From Heuristics to Experience in Progressive Pruning for LLM-based Multi-Agent Communication](https://mlsanity.com/paper/2508.11733/) (2025)

## Abstract (arXiv)

Topology-guided safeguards for LLM-based multi-agent systems (MAS) train a GNN over the inter-agent communication graph to localize risky agents and intervene on the topology---but they assume one operator can pool all labeled traces. Across organizations that assumption breaks: episodes contain private prompts, tool outputs, and proprietary workflows, and no silo alone sees the full attack distribution. We cast privacy-preserving MAS safeguarding as graph federated learning and instantiate FGLGuard: each operator fits an edge-featured graph attention detector on its own judge-labeled episode graphs and shares only model updates. The method couples a proximal local objective for non-IID clients, domain-balanced aggregation, over-refusal-constrained threshold calibration, corroborated upstream scoring, and a guarded rewrite for blocked answers. Federation is not optional: off-the-shelf transfer collapses under distribution shift (AUROC 0.51 to 0.70 only after in-domain retraining), so a deployable guard must adapt on each site's private traces. On Agent-SafetyBench, R-Judge, and AgentDojo, federated FGLGuard exceeds the in-domain centralized ceiling on all three benchmarks without pooling any data---where unsupervised anomaly guards and local-only training fail. One guard federated across four different-domain operators comes within 0.03 AUROC of multi-domain centralization, while any single-domain guard collapses on the others. Live FGLGuard cuts AgentDojo's ground-truth attack-success rate by 43% at near-unguarded utility, zero API cost, and negligible capability loss.