Minimizing Mismatch Risk: A Prototype-Based Routing Framework for Zero-shot LLM-generated Text Detection

TL;DR

This paper introduces DetectRouter, a prototype-based routing framework that dynamically selects the most suitable surrogate model for zero-shot LLM-generated text detection, significantly improving robustness and accuracy.

Contribution

It proposes a novel routing approach that learns to match surrogates to inputs, addressing the variability in detection performance due to surrogate-source alignment.

Findings

DetectRouter outperforms fixed surrogate methods on EvoBench and MAGE benchmarks.

The framework achieves consistent improvements across multiple detection criteria.

It effectively generalizes from white-box to black-box models through geometric alignment.

Abstract

Zero-shot methods detect LLM-generated text by computing statistical signatures using a surrogate model. Existing approaches typically employ a fixed surrogate for all inputs regardless of the unknown source. We systematically examine this design and find that detection performance varies substantially depending on surrogate-source alignment. We observe that while no single surrogate achieves optimal performance universally, a well-matched surrogate typically exists within a diverse pool for any given input. This finding transforms robust detection into a routing problem: selecting the most appropriate surrogate for each input. We propose DetectRouter, a prototype-based framework that learns text-detector affinity through two-stage training. The first stage constructs discriminative prototypes from white-box models; the second generalizes to black-box sources by aligning geometric distances with observed detection scores. Experiments on EvoBench and MAGE benchmarks demonstrate consistent improvements across multiple detection criteria and model families.