{
  "id": 6205,
  "url": "https://arxiv.org/abs/2604.06831v1",
  "title": "Towards Privacy-Preserving Large Language Model: Text-free Inference Through Alignment and Adaptation",
  "summary": "Current LLM-based services typically require users to submit raw text regardless of its sensitivity. While intuitive, such practice introduces substantial privacy risks, as unauthorized access may expose personal, medical, or legal information. Although prior defenses strived to mitigate these risks, they often incur substantial computational overhead and degrade model performance. To overcome this privacy-efficiency trade-off, we introduce Privacy-Preserving Fine-Tuning (PPFT), a novel training",
  "authors": "Jeongho Yoon, Chanhee Park, Yongchan Chun, Hyeonseok Moon, Heuiseok Lim",
  "category": "research",
  "topics": "safety-alignment,privacy-surveillance,healthcare",
  "orgs": null,
  "regions": null,
  "published_at": "2026-04-08T08:49:17.000Z",
  "fetched_at": "2026-07-14T16:32:20.054Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/6205",
  "original_url": "https://arxiv.org/abs/2604.06831v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}