{
  "id": 3247,
  "url": "https://arxiv.org/abs/2606.03089v2",
  "title": "Constitutional On-Policy Safe Distillation",
  "summary": "On-policy self-distillation (OPSD) has emerged as an efficient post-training paradigm by using a teacher conditioned on privileged information to provide dense token-level supervision. Prior work has shown that OPSD can collapse in verifiable reasoning tasks, but safety alignment differs in that it is guided by high-level constitutions rather than explicit target answers, making it a natural setting to revisit dense distillation. However, our pilot study show that safety OPSD still suffers from ",
  "authors": "Ming Wen, Yuxuan Liu, Kun Yang, Yunhao Feng, Zhuoer Xu, Yuhao Sun et al.",
  "category": "research",
  "topics": "regulation,safety-alignment",
  "orgs": null,
  "regions": null,
  "published_at": "2026-06-02T03:17:56.000Z",
  "fetched_at": "2026-07-14T16:30:05.532Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/3247",
  "original_url": "https://arxiv.org/abs/2606.03089v2",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}