{
  "id": 6911,
  "url": "https://arxiv.org/abs/2603.20957v3",
  "title": "Alignment Whack-a-Mole : Finetuning Activates Verbatim Recall of Copyrighted Books in Large Language Models",
  "summary": "Frontier LLM companies have repeatedly assured courts and regulators that their models do not store copies of training data. They further rely on safety alignment strategies via RLHF, system prompts, and output filters to block verbatim regurgitation of copyrighted works, and have cited the efficacy of these measures in their legal defenses against copyright infringement claims. We show that finetuning bypasses these protections: by training models to expand plot summaries into full text, a task",
  "authors": "Xinyue Liu, Niloofar Mireshghallah, Jane C. Ginsburg, Tuhin Chakrabarty",
  "category": "research",
  "topics": "regulation,safety-alignment,copyright-ip",
  "orgs": null,
  "regions": null,
  "published_at": "2026-03-21T21:46:16.000Z",
  "fetched_at": "2026-07-14T16:32:50.146Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/6911",
  "original_url": "https://arxiv.org/abs/2603.20957v3",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}