SafeX (Mixture-of-experts routing): superseded — cited as a baseline and beaten by newer methods. 2 paper(s) critique it, 1 beat it on benchmarks — #42 of 1370 most-superseded. Sub-problem: cluster led by Mixtral. Newer alternatives in the same sub-problem include PADD, MESA, PR2, RouteHijack, MASCing.

Mixture-of-experts routing

SafeX

SAFEx: Analyzing Vulnerabilities of MoE-Based LLMs via Stable Safety-critical Expert Identification

Superseded baseline#42 of 1,370 most-superseded · first seen Jun 20, 2025

Superseded — cited as a baseline and beaten by newer methods

2 papers critique it · 1 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites SafeX as a baseline.

SafeX requires computationally expensive fine-tuning and leaves routing-based safety shortcuts unexamined.

— MASCing: Configurable Mixture-of-Experts Behavior via Activation Steering Masks

“SafeX attempts to protect general reasoning by applying additive weight merging to localized safety pathways; however, this constrained intervention limits the diffusion of safety responsibility, resulting in incomplete defenses.”

— MESA: Improving MoE Safety Alignment via Decentralized Expertise

Beaten on benchmarks

Head-to-head results where a newer method reports beating SafeX. Values are copied from the source paper's tables — verify against the cited paper.

MESA beats SafeX
90.90 vs 64.00
WildJB · [Model: DeepSeek-v2-Lite]
MESA: Improving MoE Safety Alignment via Decentralized Expertise

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.