{
  "id": 14817,
  "url": "https://arxiv.org/abs/2607.26886v1",
  "title": "Hearsay: Vision-Language Medical Diagnoses Without an Image",
  "summary": "When asked to describe a medical image that was never attached, frontier vision-language models do not abstain: they confabulate a diagnosis. We show that this confabulation is not random. It is structured by who the patient is said to be. Across chest X-ray, brain MRI, and dermatology, Claude Opus-4.7, GPT-5.4, and Gemini-3.1-Pro are each queried with only a demographic descriptor and no image, and changing the descriptor systematically shifts the diagnosis returned. Claude concentrates sharply",
  "authors": "Siddharth Vohra",
  "category": "research",
  "topics": "healthcare",
  "orgs": "openai,anthropic,google",
  "regions": null,
  "published_at": "2026-07-29T13:15:23.000Z",
  "fetched_at": "2026-07-30T05:10:24.387Z",
  "source_slug": "x-arxiv-cs-ai",
  "source_name": "arXiv cs.AI",
  "source_homepage": "https://arxiv.org/list/cs.AI/recent",
  "ethics_ai_record_url": "https://ethics.ai/record/14817",
  "original_url": "https://arxiv.org/abs/2607.26886v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}