# Delegated Misalignment: How Multi-Agent Structures Amplify LLM Safety Risks

arXiv: https://arxiv.org/abs/2609.27900 · HTML: https://mlsanity.com/paper/2609.27900/

## 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 13 of its 31 references that are in the map.
- First posted on arXiv 26 Aug 2026.

## Placed in

- LLM agents and tool use / [Agent threats, guardrails and governance](https://mlsanity.com/agents/agent-threats-guardrails-and-governance/)
- LLM alignment / RLHF / preference optimisation / [Safety alignment and jailbreak attacks](https://mlsanity.com/alignment/safety-alignment-and-jailbreak-attacks/)
- LLM safety and security / [Safety alignment, refusal and guardrails](https://mlsanity.com/safety/safety-alignment-refusal-and-guardrails/)

## Builds on (most cited first, 13 of 13)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Jailbreaking Black Box Large Language Models in Twenty Queries](https://mlsanity.com/paper/2310.08419/) (2023)
- [Identifying the Risks of LM Agents with an LM-Emulated Sandbox](https://mlsanity.com/paper/2309.15817/) (2023)
- [AgentHarm: A Benchmark for Measuring Harmfulness of LLM Agents](https://mlsanity.com/paper/2410.09024/) (2024)
- [Multi-Agent Collaboration: Harnessing the Power of Intelligent LLM Agents](https://mlsanity.com/paper/2306.03314/) (2023)
- [Jailbreak Vision Language Models via Bi-Modal Adversarial Prompt](https://mlsanity.com/paper/2406.04031/) (2024)
- [Here Comes The AI Worm: Unleashing Zero-click Worms that Target GenAI-Powered Applications](https://mlsanity.com/paper/2403.02817/) (2024)
- [GameGPT: Multi-agent Collaborative Framework for Game Development](https://mlsanity.com/paper/2310.08067/) (2023)
- [SecureWebArena: A Holistic Security Evaluation Benchmark for LVLM-based Web Agents](https://mlsanity.com/paper/2510.10073/) (2025)
- [Self-Organizing Agent Network for LLM-based Workflow Automation](https://mlsanity.com/paper/2508.13732/) (2025)
- [AgentVisor: Defending LLM Agents Against Prompt Injection via Semantic Virtualization](https://mlsanity.com/paper/2604.24118/) (2026)
- [Evolving Deception: When Agents Evolve, Deception Wins](https://mlsanity.com/paper/2603.05872/) (2026)

## Abstract (arXiv)

Large language models (LLMs) are increasingly deployed in multi-agent systems where a principal agent decomposes tasks and delegates them to subordinate agents that may invoke external tools. Safety alignment, however, is still evaluated almost exclusively under a single-agent threat model, treating safety as a property of the individual LLM. We show that this assumption breaks down: individual safety alignment fails to transfer to multi-agent settings. Two failure mechanisms emerge under delegation: responsibility diffusion on the principal side and role-bias compliance on the subordinate side, jointly converting language-level refusal into actionable harm. We refer to this phenomenon as textitdelegated misalignment and study it through a three-condition protocol across 6 frontier LLMs on 49 hazardous tasks. Delegation amplifies end-to-end harm substantially: DeepSeek-V3.2's full-execution rate rises from 30.6% to 77.6% once delegation is introduced, and the same model behaves very differently across roles (GPT-5: 22.5% as a single agent vs. 61.2% as a subordinate). Ablations further show that standard single-layer defenses each fail on their own and can even backfire. We call on the community to move beyond per-model alignment and toward composite safety mechanisms before multi-agent LLM systems are deployed at scale.