{
  "id": 16929,
  "url": "https://arxiv.org/abs/2608.04896v1",
  "title": "When Shared Rollouts Fail in Defensive Driving Evaluation: A NAVSIM Score Basis Audit",
  "summary": "Defensive driving scores are useful only when they preserve distinctions between policies that observe surrounding actors and those that do not. Re-simulation benchmarks may use reference-conditioned forgiveness, under which an agent receives credit when the logged human reference fails a compliance channel. When agent and reference share an unstable rollout transformation, this rule can propagate shared reference failures into broad compliance credit. We audit this risk in NAVSIM v2.2 original",
  "authors": "Ziang Wei, Minjun Yu, Zheyuan Lai, Mingjie Pang, Wei Li",
  "category": "research",
  "topics": "regulation,agents-autonomy,transparency",
  "orgs": null,
  "regions": null,
  "published_at": "2026-08-05T14:22:23.000Z",
  "fetched_at": "2026-08-06T05:10:11.148Z",
  "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/16929",
  "original_url": "https://arxiv.org/abs/2608.04896v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}