{
  "id": 3316,
  "url": "https://arxiv.org/abs/2606.09864v1",
  "title": "Alignment Collapse Under KV Cache Quantization: Diagnosis and Mitigation",
  "summary": "Key-value (KV) cache quantization is widely used to reduce Large Language Model (LLM) inference memory, yet existing evaluations solely focus on measuring perplexity and accuracy without assessing the safety impact. In this study, we explore alignment preservation under KV cache quantization. Across eleven instruction-tuned models (3.8B-72B) and five benchmarks (1,894 prompts), we find that low-bit quantization can silently destroy safety alignment: Mistral-7B loses 15.2% of its refusals at only",
  "authors": "Bruce Changlong Xu, Adarsh Kumarappan, Mu Zhou",
  "category": "research",
  "topics": "safety-alignment,healthcare",
  "orgs": "mistral",
  "regions": null,
  "published_at": "2026-06-01T02:02:20.000Z",
  "fetched_at": "2026-07-14T16:30:09.961Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/3316",
  "original_url": "https://arxiv.org/abs/2606.09864v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}