Work

Reasoning red teaming in healthcare not all paths to a desired outcome are desirable

Year: 2025

Type: article

Abstract: Chang et al. showed that large language models can produce unsafe or biased outputs even when superficially accurate. We highlight that LLMs can hide harmful reasoning if only final responses are red-teamed. Monitoring intermediate inference steps, especially in ethically charged clinical scenarios, can reveal manipulative or unethical thought processes. We propose systematic testing of ethically sensitive prompts and thorough chain-of-thought analysis to ensure safe, trustworthy deployment in healthcare.

Source: npj Digital Medicine

DOI: 10.1038/s41746-025-02104-0

Authors: Vera Sorin, Panagiotis Korfiatis, Girish N. Nadkarni, Eyal Klang

Institutions: Mayo Clinic, Mount Sinai Hospital, Icahn School of Medicine at Mount Sinai, Mount Sinai Health System

Language: English

Topic: Artificial Intelligence in Healthcare and Education

Subfield: Health Informatics

Field: Medicine

Domain: Health Sciences

SDG: Good health and well-being

Open Access status: gold

Estimated APC paid: $4,090