paper-with-me

Papers

Shallow Robustness, Deep Vulnerabilities: Multi-Turn Evaluation of Medical LLMs

2025-10-14 · Blazej Manczak, Eric Lin, Francisco Eiras, James O' Neill, Vaikkunth Mugunthan arxiv

Large language models (LLMs) are rapidly transitioning into medical clinical use, yet their reliability under realistic, multi-turn interactions remains poorly understood. Existing evaluation frameworks typically assess single-turn question answering under idealized conditions, overlooking the complexities of medical consultations where conflicting input, misleading context, and authority influence are common. We introduce MedQA-Followup, a framework for systematically evaluating multi-turn robustness in medical question answering. Our approach distinguishes between shallow robustness (resisting misleading initial context) and deep robustness (maintaining accuracy when answers are challenged across turns), while also introducing an indirect-direct axis that separates contextual framing (indirect) from explicit suggestion (direct). Using controlled interventions on the MedQA dataset, we evaluate five state-of-the-art LLMs and find that while models perform reasonably well under shallow perturbations, they exhibit severe vulnerabilities in multi-turn settings, with accuracy dropping from 91.2% to as low as 13.5% for Claude Sonnet 4. Counterintuitively, indirect, context-based interventions are often more harmful than direct suggestions, yielding larger accuracy drops across models and exposing a significant vulnerability for clinical deployment. Further compounding analyses reveal model differences, with some showing additional performance drops under repeated interventions while others partially recovering or even improving. These findings highlight multi-turn robustness as a critical but underexplored dimension for safe and reliable deployment of medical LLMs.

📄 PDF Abstract BibTeX arXiv:2510.12255

Code (0)

등록된 구현이 없습니다.

Tasks

Question Answering

Results from the Paper

RankTaskDatasetModelMetrics
#1 Question Answering MedQA MedQA-Followup Accuracy: 91.2

Similar Papers 제목 키워드 기반

MOCHA: Are Code Language Models Robust Against Multi-Turn Malicious Coding Prompts?

2025-07-25 · Muntasir Wahed, Xiaona Zhou, Kiet A. Nguyen, Tianjiao Yu 외 arxiv

Recent advancements in Large Language Models (LLMs) have significantly enhanced their code generation capabilities. However, their robustness against adversarial misuse, particularly through multi-turn malicious coding p…

Code Generation

Transient Turn Injection: Exposing Stateless Multi-Turn Vulnerabilities in Large Language Models

2026-04-23 · Naheed Rayhan, Sohely Jahan arxiv

Large language models (LLMs) are increasingly integrated into sensitive workflows, raising the stakes for adversarial robustness and safety. This paper introduces Transient Turn Injection(TTI), a new multi-turn attack te…

Adversarial Robustness

Evaluating False Alarm and Missing Attacks in CAN IDS

2026-02-02 · Nirab Hossain, Pablo Moriano arxiv

Modern vehicles rely on electronic control units (ECUs) interconnected through the Controller Area Network (CAN), making in-vehicle communication a critical security concern. Machine learning (ML)-based intrusion detecti…

Adversarial RobustnessIntrusion Detection

Safety Alignment Should Be Made More Than Just a Few Tokens Deep

2024-06-10 · Xiangyu Qi, Ashwinee Panda, Kaifeng Lyu, Xiao Ma 외

The safety alignment of current Large Language Models (LLMs) is vulnerable. Relatively simple attacks, or even benign fine-tuning, can jailbreak aligned models. We argue that many of these vulnerabilities are related to …

Safety Alignment

Is Length Really A Liability? An Evaluation of Multi-turn LLM Conversations using BoolQ

2026-01-23 · Karl Neergaard, Le Qiu, Emmanuele Chersoni arxiv

Single-prompt evaluations dominate current LLM benchmarking, yet they fail to capture the conversational dynamics where real-world harm occurs. In this study, we examined whether conversation length affects response vera…