# When Local Monitors Miss Compositional Harm: Diagnosing Distributed Backdoors in Multi-Agent Systems

arXiv: https://arxiv.org/abs/2607.11751 · HTML: https://mlsanity.com/paper/2607.11751/

## Key facts

- Cited by 2 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 27 of its 41 references that are in the map.
- First posted on arXiv 13 Jul 2026.

## Placed in

- LLM agents and tool use / [Agent threats, guardrails and governance](https://mlsanity.com/agents/agent-threats-guardrails-and-governance/)
- LLM safety and security / [Prompt injection and agent security](https://mlsanity.com/safety/prompt-injection-and-agent-security/)

## Builds on (most cited first, 27 of 27)

- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Generative Agents: Interactive Simulacra of Human Behavior](https://mlsanity.com/paper/2304.03442/) (2023)
- [Universal and Transferable Adversarial Attacks on Aligned Language Models](https://mlsanity.com/paper/2307.15043/) (2023)
- [Shortcut Learning in Deep Neural Networks](https://mlsanity.com/paper/2004.07780/) (2020)
- [Jailbroken: How Does LLM Safety Training Fail?](https://mlsanity.com/paper/2307.02483/) (2023)
- [Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection](https://mlsanity.com/paper/2302.12173/) (2023)
- [Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations](https://mlsanity.com/paper/2312.06674/) (2023)
- [Right for the Wrong Reasons: Diagnosing Syntactic Heuristics in Natural Language Inference](https://mlsanity.com/paper/1902.01007/) (2019)
- [Ignore Previous Prompt: Attack Techniques For Language Models](https://mlsanity.com/paper/2211.09527/) (2022)
- [Annotation Artifacts in Natural Language Inference Data](https://mlsanity.com/paper/1803.02324/) (2018)
- [R-Judge: Benchmarking Safety Risk Awareness for LLM Agents](https://mlsanity.com/paper/2401.10019/) (2024)
- [DrAttack: Prompt Decomposition and Reconstruction Makes Powerful LLM Jailbreakers](https://mlsanity.com/paper/2402.16914/) (2024)
- [OpenAgentSafety: A Comprehensive Framework for Evaluating Real-World AI Agent Safety](https://mlsanity.com/paper/2507.06134/) (2025)
- [From Prompt Injections to Protocol Exploits: Threats in LLM-Powered AI Agents Workflows](https://mlsanity.com/paper/2506.23260/) (2025)
- [LLM Censorship: A Machine Learning Challenge or a Computer Security Problem?](https://mlsanity.com/paper/2307.10719/) (2023)
- [AgentArmor: Enforcing Program Analysis on Agent Runtime Trace to Defend Against Prompt Injection](https://mlsanity.com/paper/2508.01249/) (2025)
- [SkillTrojan: Backdoor Attacks on Skill-Based Agent Systems](https://mlsanity.com/paper/2604.06811/) (2026)
- [DemonAgent: Dynamically Encrypted Multi-Backdoor Implantation Attack on LLM-based Agent](https://mlsanity.com/paper/2502.12575/) (2025)
- [Lying with Truths: Open-Channel Multi-Agent Collusion for Belief Manipulation via Generative Montage](https://mlsanity.com/paper/2601.01685/) (2026)
- [BackdoorAgent: A Unified Framework for Backdoor Attacks on LLM-based Agents](https://mlsanity.com/paper/2601.04566/) (2026)
- [Ghost in the Agent: Redefining Information Flow Tracking for LLM Agents](https://mlsanity.com/paper/2604.23374/) (2026)
- [PeerGuard: Defending Multi-Agent Systems Against Backdoor Attacks Through Mutual Reasoning](https://mlsanity.com/paper/2505.11642/) (2025)
- [When Benchmarks Lie: Evaluating Malicious Prompt Classifiers Under True Distribution Shift](https://mlsanity.com/paper/2602.14161/) (2026)
- [Your Agent Can Defend Itself against Backdoor Attacks](https://mlsanity.com/paper/2506.08336/) (2025)
- [SafetyDrift: Predicting When AI Agents Cross the Line Before They Actually Do](https://mlsanity.com/paper/2603.27148/) (2026)
- [Most LLM Conformity Needs No Speaker: Measuring the Speaker-Free Floor in Peer-Pressure Benchmarks](https://mlsanity.com/paper/2607.05545/) (2026)
- [When Better Codebooks Are Not Enough: Predictive Performance and Behavioral Reliability in LLM Political Event Coding](https://mlsanity.com/paper/2606.06781/) (2026)

## Built on by (most cited first, 2 of 2)

- [Silence Is Endorsement: Verification-Status Laundering in LLM Agent Pipelines](https://mlsanity.com/paper/2609.20211/) (2026)
- [Early Detection of Distributed Backdoors in Multi-Agent LLM Systems: A Characterization Study](https://mlsanity.com/paper/2607.24893/) (2026)

## Abstract (arXiv)

As multi-agent, tool-using LLM systems are deployed, a common safety net is a runtime monitor that checks each message, tool call, or step on its own. We show this net has a fundamental hole. A distributed backdoor splits a harmful payload across agents, so every local check passes while the assembled object is the attack. The monitor can be right on every step and still miss the attack. The problem is not splitting itself: split fragments can still leak suspicious tokens or provenance edges. The hard case is local benignness. No fragment carries the harm, and what is left looks like ordinary benign traffic. We formalize this as an observability boundary: a monitor catches only what its view can tell apart from benign traffic. We prove that once the fragments look benign in the monitored view, no detector on that view can catch them, however strong it is. Across a controlled testbed, an external benchmark, and end-to-end agent runs, local monitors lose the signal exactly as local evidence disappears, and it returns only when the monitor sees the assembled object. A monitor trained only on benign traffic recovers the attack's code structure across held-out encodings (0.874 mean AUROC). A decoded-view gate, given the encoding family, blocks every tested attack. But seeing more is not enough: full-trace monitors and decoders still fail unless they reach the representation where the payload is exposed. Local safety is not global safety when harm is compositional, and the open problem is finding that representation.