Medical AI Risk

Medical AI risk refers to the potential for artificial intelligence systems used in healthcare, such as diagnostic imaging tools, clinical decision support software, and patient triage systems, to produce errors that harm patients or expose providers to liability. Failures can include missed diagnoses, biased performance across patient populations, and recommendations that clinicians follow without adequate scrutiny. For compliance officers and general counsel, this tag tracks regulatory actions, FDA oversight of AI-enabled medical devices, malpractice exposure, and privacy obligations tied to health data used in model training.

1 item

ResearchGlobal2026-10-26

Old CNN Beats GPT-6 Astra and Claude Fable 5.1 on Skin Cancer Diagnosis

A study by dermatologist Dr. Magnus Lynch tested GPT-6 Astra and Claude Fable 5.1 against a ten-year-old specialized image classifier on 200 biopsy-confirmed skin lesions. The specialized model reached 62% diagnostic accuracy; both frontier large language models (LLMs) reached only 42%. The finding directly challenges procurement assumptions that general-purpose frontier AI is fit for high-stakes clinical tasks.

GPT-6 Astra Claude Fable 5.1 OpenAI Anthropic clinical AI model evaluation medical AI risk AI procurement