{
  "id": 935,
  "url": "https://arxiv.org/abs/2606.17257v1",
  "title": "Pulling The REINS: Training-Free Safety Alignment of Video Diffusion Models via Representation Steering",
  "summary": "Open-weight video diffusion models can generate photorealistic unsafe content, from violence to misinformation, yet existing defenses either require expensive safety fine-tuning that degrades general capability, or apply external filters that are trivially bypassed by adversarial prompts. We present REINS (REpresentation-space INference-time Safety steering), a training-free method that aligns video diffusion models at inference time by steering their internal representations toward safe generat",
  "authors": "Rohit Kundu, Arindam Dutta, Sarosij Bose, Athula Balachandran, Amit K. Roy-Chowdhury",
  "category": "research",
  "topics": "safety-alignment,misinformation",
  "orgs": null,
  "regions": null,
  "published_at": "2026-06-15T20:03:09.000Z",
  "fetched_at": "2026-07-14T14:14:54.532Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/935",
  "original_url": "https://arxiv.org/abs/2606.17257v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}