# Bilevel Optimization for Covert Memory Tampering in Heterogeneous Multi-Agent Architectures (XAMT)

arXiv: https://arxiv.org/abs/2512.15790 · HTML: https://mlsanity.com/paper/2512.15790/

## Key facts

- Cited by 1 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 11 of its 35 references that are in the map.
- First posted on arXiv 15 Dec 2025.

## 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, 11 of 11)

- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [Towards Evaluating the Robustness of Neural Networks](https://mlsanity.com/paper/1608.04644/) (2016)
- [The StarCraft Multi-Agent Challenge](https://mlsanity.com/paper/1902.04043/) (2019)
- [AgentPoison: Red-teaming LLM Agents via Poisoning Memory or Knowledge Bases](https://mlsanity.com/paper/2407.12784/) (2024)
- [Witches' Brew: Industrial Scale Data Poisoning via Gradient Matching](https://mlsanity.com/paper/2009.02276/) (2020)
- [Machine Against the RAG: Jamming Retrieval-Augmented Generation with Blocker Documents](https://mlsanity.com/paper/2406.05870/) (2024)
- [On Penalty-based Bilevel Gradient Descent Method](https://mlsanity.com/paper/2302.05185/) (2023)
- [SafeRAG: Benchmarking Security in Retrieval-Augmented Generation of Large Language Model](https://mlsanity.com/paper/2501.18636/) (2025)
- [CPA-RAG:Covert Poisoning Attacks on Retrieval-Augmented Generation in Large Language Models](https://mlsanity.com/paper/2505.19864/) (2025)
- [Implicit Poisoning Attacks in Two-Agent Reinforcement Learning: Adversarial Policies for Training-Time Attacks](https://mlsanity.com/paper/2302.13851/) (2023)
- [Self-healing networks: redundancy and structure](https://mlsanity.com/paper/1305.3450/) (2013)

## Built on by (most cited first, 1 of 1)

- [Autonomous Research Agents: A Survey of AI Scientists and the Verification Gap](https://mlsanity.com/paper/2608.05179/) (2026)

## Abstract (arXiv)

The increasing operational reliance on complex Multi-Agent Systems (MAS) across safety-critical domains necessitates rigorous adversarial robustness assessment. Modern MAS are inherently heterogeneous, integrating conventional Multi-Agent Reinforcement Learning (MARL) with emerging Large Language Model (LLM) agent architectures utilizing Retrieval-Augmented Generation (RAG). A critical shared vulnerability is reliance on centralized memory components: the shared Experience Replay (ER) buffer in MARL and the external Knowledge Base (K) in RAG agents. This paper proposes XAMT (Bilevel Optimization for Covert Memory Tampering in Heterogeneous Multi-Agent Architectures), a novel framework that formalizes attack generation as a bilevel optimization problem. The Upper Level minimizes perturbation magnitude (delta) to enforce covertness while maximizing system behavior divergence toward an adversary-defined target (Lower Level). We provide rigorous mathematical instantiations for CTDE MARL algorithms and RAG-based LLM agents, demonstrating that bilevel optimization uniquely crafts stealthy, minimal-perturbation poisons evading detection heuristics. Comprehensive experimental protocols utilize SMAC and SafeRAG benchmarks to quantify effectiveness at sub-percent poison rates (less than or equal to 1 percent in MARL, less than or equal to 0.1 percent in RAG). XAMT defines a new unified class of training-time threats essential for developing intrinsically secure MAS, with implications for trust, formal verification, and defensive strategies prioritizing intrinsic safety over perimeter-based detection.