Safety-Oriented Routing Analysis of Mixtral MoE Under Benign and Harmful Prompts

2026-05-22

AI Summary

This study investigates how the routing mechanism of the Mixtral 8x7B-Instruct model influences safety outcomes in response to both benign and harmful prompts. By jointly analyzing expert activation frequencies and router gating gradients—and integrating targeted expert suppression with cross-group expert categorization—the work reveals, for the first time, the deep dependency and distributed nature of safety-related routing decisions. The findings demonstrate that safety-critical experts are broadly dispersed yet concentrated in specific layers; moreover, selectively suppressing experts identified via gradient-based importance significantly reduces restricted responses while inducing fewer side effects, thereby overcoming the limitations of single-metric analyses.

Abstract

Sparse mixture-of-experts (MoE) language models activate only a small subset of parameters for each token, making router behavior a central part of model computation. This paper studies routing behavior of Mixtral 8x7B-Instruct under benign and harmful prompts using two complementary signals: activation-based routing scores derived from expert selection frequencies and gradient-based scores derived from router-gate sensitivities. We analyze expert- and layer-level routing behavior and conduct expert-suppression interventions. The results show that activation-based expert usage is broad and long-tailed, whereas gradient-based importance is concentrated. At expert level, benign and harmful prompt groups remain close under both signals with modest separation. At layer level, activation-based routing is most selective around layers 8-15, while gradient-based importance is concentrated in final layers. Expert classification shows most experts are shared across benign and harmful prompts, though a limited subset shows clear group preference. Top-ranked expert sets show stronger benign-malicious overlap under gradient scores than activation scores, suggesting concentration on a common late-layer expert set. In intervention experiments, suppressing top five benign-dominant experts from activation scores reduces restricted responses from 24 to 14 over 100 prompts, while suppressing gradient-derived experts reduces them from 34 to 22 with fewer unintended reversals. Overall, safety-relevant routing in Mixtral is subtle, depth-dependent, and distributed rather than dominated by a fixed set of experts.

Problem

• mixture-of-experts
• safety
• routing behavior
• harmful prompts
• language models

Innovation

• mixture-of-experts
• routing analysis
• safety alignment
• gradient-based importance
• expert suppression