{
  "id": 4545,
  "url": "https://arxiv.org/abs/2605.10806v1",
  "title": "PhyGround: Benchmarking Physical Reasoning in Generative World Models",
  "summary": "Generative world models are increasingly used for video generation, where learned simulators are expected to capture the physical rules that govern real-world dynamics. However, evaluating whether generated videos actually follow these rules remains challenging. Existing physics-focused video benchmarks have made important progress, but they still face three key challenges, including the coarse evaluation frameworks that hide law-specific failures, response biases and fatigue that undermine the ",
  "authors": "Juyi Lin, Arash Akbari, Yumei He, Lin Zhao, Haichao Zhang, Arman Akbari et al.",
  "category": "research",
  "topics": "regulation",
  "orgs": null,
  "regions": null,
  "published_at": "2026-05-11T16:30:51.000Z",
  "fetched_at": "2026-07-14T16:31:03.581Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/4545",
  "original_url": "https://arxiv.org/abs/2605.10806v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}