Article

LLMs Can Do Medical Harm: Stress-Testing Clinical Decisions Under Social Pressure

2025-11-27

Abstract excerpt

<h4>Background</h4> Large language models (LLMs) are entering clinical workflows, yet their effect on clinical decisions and potential for harm are uncertain. <h4>Methods</h4> We measured harmful decision output from an ensemble of 20 LLMs across >10 million clinical scenarios with safety or ethical dilemmas. Each case was shown under a neutral control and six Milgram-style social-pressure conditions, with or wi...

Topics

Open a Topic to create a Post that cites this publication.

Identifiers and source

Literature Corpus work: 1db7ab18-b423-5b01-ad79-c5996625388b
DOI: 10.1101/2025.11.25.25340972

Open publication

Related research

Semantic proximity does not establish scientific evidence.

Click a neighbor to travelStep 1 · 12 closest

LLMs Can Do Medical Harm: Stress-Testing Clinical Decisions Under Social PressureDOI 10.1101/2025.11.25.25340972Read

Large Language Models Are Highly Vulnerable to Adversarial Hallucination Attacks in Clinical Decision Support: A Multi-Model Assurance AnalysisSemantic proximity Read

Simulating Lay Health-Seeking Behavior with LLM Personas and Illness Vignettes: Reproducibility, Prompt Sensitivity, and Slice DependenceSemantic proximity Read

Impact of Patient Communication Style on Agentic AI-Generated Clinical Advice in E-MedicineSemantic proximity Read

LLM Reasoning Does Not Protect Against Clinical Cognitive Biases - An Evaluation Using BiasMedQASemantic proximity Read

Evaluation of four large language models on complex, infectious disease case scenariosSemantic proximity Read

When Algorithms Meet Ethics: Systematic Evidence of Framing Effects in LLM Organizational Decision-MakingSemantic proximity Read

Beyond Benchmarks: Dynamic, Automatic And Systematic Red-Teaming Agents For Trustworthy Medical Language ModelsSemantic proximity Read

Evaluating anti-LGBTQIA+ medical bias in large language modelsSemantic proximity Read

A statistical framework for evaluating the repeatability and reproducibility of large language modelsSemantic proximity Read

Simulating Lay Health-Seeking Behavior with LLM Personas and Illness Vignettes: Reproducibility, Prompt Sensitivity, and Slice DependenceSemantic proximity Read

Clinical Safety of AI-Generated Antibiotic Prescribing Advice: Guideline Adherence and Misinformation Risk Among Large Language ModelsSemantic proximity Read

Impact of LLM Assistance on Physician Decision-Making: A Multi-Country Randomized Controlled Trial <sup>∗</sup>Semantic proximity Read