{
  "id": 20547097,
  "title": "Reading Scene States: Uncovering Safety Failures in Vision-Language Models",
  "url": "https://urgent.news/2026/11/17/reading-scene-states-uncovering-safety-failures-in-vision-language",
  "topic": "ai",
  "section": "AI",
  "published": "2026-11-17T18:48:02.000Z",
  "source": {
    "name": "arXiv cs.AI",
    "slug": "arxiv-cs-ai",
    "url": "https://arxiv.org/abs/2611.23206v1"
  },
  "original_language": "en",
  "account": null,
  "summary": "Vision-language models (VLMs) can provide harmful assistance by interpreting visual events, even without an explicit request for the underlying harmful objective. We identify this \\textbf{scene-state safety vulnerability} and introduce \\textbf{SceneSafetyBench}, comprising 375 image--question pairs spanning Initial, Medium, and Ultimate scene states and five visual presentations. Across eight…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}