Route to Rome Attack: Directing LLM Routers to Expensive Models via Adversarial Suffix Optimization

ACL ARR 2026 January Submission4710 Authors

Keywords: adversarial attacks, router attack, large language models, LLM routing

Abstract: Cost-aware routing dynamically dispatches user queries to models of varying capability to balance performance and inference cost. However, the routing strategy introduces new security concern that adversaries may manipulate router to consistently select expensive high-capability models. Existing routing attacks depend either on white-box access or heuristic prompts, rendering them ineffective in real-world black-box scenarios. In this work, we propose R$^2$A which aims to mislead black-box LLM routers to expensive models via adversarial suffix optimization. Specifically, R$^2$A deploys a hybrid ensemble surrogate router to mimic the black-box router. A suffix optimization algorithm is further adapted for the ensemble-based surrogate. Extensive experiments on multiple open-source and commercial routing systems demonstrate that R$^2$A significantly increases the routing rate to expensive models on queries of different distributions. Code and examples: https://anonymous.4open.science/r/Anonymous_code-692E.

Paper Type: Long

Research Area: Ethics, Bias, and Fairness

Research Area Keywords: ethical considerations in NLP applications

Contribution Types: Model analysis & interpretability, NLP engineering experiment, Publicly available software and/or pre-trained models

Languages Studied: English

Submission Number: 4710