MultiTurnPSB: Evaluating Multi-Turn Jailbreak Attacks an dClassifier-Based Defenses for Medical AI Safety
Patient-facing medical chatbots are commonly evaluated on single-turn prompts, yet real users push back after refusals, add urgency, and invoke authority. We introduce MultiTurnPSB, a four-turn adversarial extension of PatientSafetyBench, and evaluate GPT-4.1-mini under fixed template, template-adaptive, and live adversarial attacks. Unsafe responses rise from 35% to nearly 80% by Turn 4 under live attack. Under the same adversary, GPT-4.1-mini and Claude Sonnet 4.5 are statistically indistingui
Record details
Published: 30 May 2026
Source: arXiv
Category: Research
Topics: Safety & alignment · Healthcare
Retrieved: 14 July 2026
Related evidence
These records share source-supplied organisations, an exact publisher byline, automatic topics or regions. The reason is shown on every link; related does not mean supporting, agreeing with or verifying this record.
Large language models provide unsafe answers to patient-posed medical questions
OpenAlex · 13 February 2026
How Well Do Models Follow Their Constitutions?
arXiv · 22 May 2026
Can AI Make Conflicts Worse? An Alignment Failure in LLM Deployment Across Conflict Contexts
arXiv · 21 May 2026
Geopolitical alignment: Endorsement effects in large language models
arXiv · 10 July 2026
When AI Tells You What You Want to Hear: Sycophantic Behavior of Large Language Models in Dementia Care Settings
arXiv · 13 April 2026
Socioeconomic Inference in LLM Medical Triage: Same Symptoms, Different ZIP Code
arXiv cs.CY · 28 July 2026
How to cite this record
ethics.ai (30 May 2026), “MultiTurnPSB: Evaluating Multi-Turn Jailbreak Attacks an dClassifier-Based Defenses for Medical AI Safety,” evidence record 3391, https://ethics.ai/record/3391 (originally published by arXiv).
Use and limitations
This page is a stable index and citation surface for a source record. ethics.ai did not author the underlying report and has not independently verified every claim. Automatic topics may be imperfect. For consequential use, quote and cite the original publisher.