Cover Image

BadMoE: Backdooring Mixture-of-Experts LLMs via Optimizing Routing Triggers and Infecting Dormant Experts

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Wang, Qingyue, Pang, Qi, Lin, Xixun, Wang, Shuai, Wu, Daoyuan
Format:	Preprint
Published:	2025
Subjects:	Cryptography and Security Artificial Intelligence
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Who Speaks for the Trigger? Dynamic Expert Routing in Backdoored Mixture-of-Experts Transformers
by: Zhao, Xin, et al.
Published: (2025)

BadPatches: Routing-aware Backdoor Attacks on Vision Mixture of Experts
by: Chan, Cedric, et al.
Published: (2025)

CryptoMoE: Privacy-Preserving and Scalable Mixture of Experts Inference via Balanced Expert Routing
by: Zhou, Yifan, et al.
Published: (2025)

Misrouter: Exploiting Routing Mechanisms for Input-Only Attacks on Mixture-of-Experts LLMs
by: Fei, Zekun, et al.
Published: (2026)

Condition-Triggered Cryptographic Asset Control via Dormant Authorization Paths
by: Wang, Jian Sheng
Published: (2026)

Large Language Lobotomy: Jailbreaking Mixture-of-Experts via Expert Silencing
by: Lintelo, Jona te, et al.
Published: (2026)

Backdoor Contrastive Learning via Bi-level Trigger Optimization
by: Sun, Weiyu, et al.
Published: (2024)

GateBreaker: Gate-Guided Attacks on Mixture-of-Expert LLMs
by: Wu, Lichao, et al.
Published: (2025)

MoPE: A Mixture of Password Experts for Improving Password Guessing
by: Duan, Mingjian, et al.
Published: (2025)

RASA: Routing-Aware Safety Alignment for Mixture-of-Experts Models
by: Liang, Jiacheng, et al.
Published: (2026)

Sparse Models, Sparse Safety: Unsafe Routes in Mixture-of-Experts LLMs
by: Jiang, Yukun, et al.
Published: (2026)

MoEcho: Exploiting Side-Channel Attacks to Compromise User Privacy in Mixture-of-Experts LLMs
by: Ding, Ruyi, et al.
Published: (2025)

TooBadRL: Trigger Optimization to Boost Effectiveness of Backdoor Attacks on Deep Reinforcement Learning
by: Zhang, Mingxuan, et al.
Published: (2025)

Differentiation-Based Extraction of Proprietary Data from Fine-Tuned LLMs
by: Li, Zongjie, et al.
Published: (2025)

MASCing: Configurable Mixture-of-Experts Behavior via Activation Steering Masks
by: Lintelo, Jona te, et al.
Published: (2026)

LLMs Can Defend Themselves Against Jailbreaking in a Practical Manner: A Vision Paper
by: Wu, Daoyuan, et al.
Published: (2024)

Routing-Aware Explanations for Mixture of Experts Graph Models in Malware Detection
by: Shokouhinejad, Hossein, et al.
Published: (2026)

BadLLM-TG: A Backdoor Defender powered by LLM Trigger Generator
by: Zhang, Ruyi, et al.
Published: (2026)

Trusted Weights, Treacherous Optimizations? Optimization-Triggered Backdoor Attacks on LLMs
by: Wang, Yifei, et al.
Published: (2026)

Backdoors in Code Summarizers: How Bad Is It?
by: Wang, Chenyu, et al.
Published: (2025)

Mixture of Robust Experts (MoRE):A Robust Denoising Method towards multiple perturbations
by: Cheng, Hao, et al.
Published: (2021)

Isolate Trigger: Detecting and Eliminating Adaptive Backdoor Attacks
by: Sun, Chengrui, et al.
Published: (2025)

Buffer Overflow in Mixture of Experts
by: Hayes, Jamie, et al.
Published: (2024)

Toward Efficient Inference Attacks: Shadow Model Sharing via Mixture-of-Experts
by: Bai, Li, et al.
Published: (2025)

SAFEx: Analyzing Vulnerabilities of MoE-Based LLMs via Stable Safety-critical Expert Identification
by: Lai, Zhenglin, et al.
Published: (2025)

Meme Trojan: Backdoor Attacks Against Hateful Meme Detection via Cross-Modal Triggers
by: Wang, Ruofei, et al.
Published: (2024)

BadTemplate: A Training-Free Backdoor Attack via Chat Template Against Large Language Models
by: Wang, Zihan, et al.
Published: (2026)

Tighter Risk Bounds for Mixtures of Experts
by: Akretche, Wissam, et al.
Published: (2024)

Revisiting Training-Inference Trigger Intensity in Backdoor Attacks
by: Lin, Chenhao, et al.
Published: (2025)

BadFair: Backdoored Fairness Attacks with Group-conditioned Triggers
by: Xue, Jiaqi, et al.
Published: (2024)

RouteScan: A Non-Intrusive Approach to Auditing MoE LLMs Safety via Expert Routing Telemetry
by: Lv, Bo, et al.
Published: (2026)

BadDLM: Backdooring Diffusion Language Models with Diverse Targets
by: Zhai, Shengfang, et al.
Published: (2026)

A Practical Trigger-Free Backdoor Attack on Neural Networks
by: Wang, Jiahao, et al.
Published: (2024)

BadRSSD: Backdoor Attacks on Regularized Self-Supervised Diffusion Models
by: Wang, Jiayao, et al.
Published: (2026)

BadTime: An Effective Backdoor Attack on Multivariate Long-Term Time Series Forecasting
by: Xiang, Kunlan, et al.
Published: (2025)

TrafficMoE: Heterogeneity-aware Mixture of Experts for Encrypted Traffic Classification
by: He, Qing, et al.
Published: (2026)

Differentially Private Training of Mixture of Experts Models
by: Tholoniat, Pierre, et al.
Published: (2024)

Trigger Where It Hurts: Unveiling Hidden Backdoors through Sensitivity with Sensitron
by: Zhao, Gejian, et al.
Published: (2025)

Enhancing Physical Layer Communication Security through Generative AI with Mixture of Experts
by: Zhao, Changyuan, et al.
Published: (2024)

BadEdit: Backdooring large language models by model editing
by: Li, Yanzhou, et al.
Published: (2024)