{
  "id": 273,
  "url": "https://arxiv.org/abs/2607.03386v1",
  "title": "When Aggregate Alignment Misleads: Auditing Policy Repair Without Per-State Expert Actions",
  "summary": "Agentic AI systems are increasingly used to edit, refine, and repair decision policies, but evaluating these edits is difficult when per-state expert action labels are unavailable. We study this problem in a hotel-pricing simulator where an agentic policy editor receives only region-level diagnostic feedback: summaries of how its price distribution differs from a benchmark policy across time, inventory, and market regions. The editor cannot observe benchmark actions, benchmark source code, rewar",
  "authors": "Peiying Zhu, Sidi Chang",
  "category": "research",
  "topics": "regulation,safety-alignment,healthcare,agents-autonomy",
  "orgs": null,
  "regions": null,
  "published_at": "2026-07-03T14:42:01.000Z",
  "fetched_at": "2026-07-14T14:14:24.248Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/273",
  "original_url": "https://arxiv.org/abs/2607.03386v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}