{
  "id": 3654,
  "url": "https://arxiv.org/abs/2605.27355v2",
  "title": "Alignment Tampering: How Reinforcement Learning from Human Feedback Is Exploited to Optimize Misaligned Biases",
  "summary": "Reinforcement Learning from Human Feedback (RLHF) is the standard method to align Large Language Models (LLMs) with human preferences. In this work, we introduce alignment tampering, a potential vulnerability where the LLM undergoing alignment influences the preference dataset, causing RLHF to amplify undesired behaviors. This arises from core limitations of RLHF: (1) preference datasets are constructed from the LLM's own outputs, allowing it to influence them, and (2) pairwise comparisons only ",
  "authors": "Dongyoon Hahm, Dylan Hadfield-Menell, Kimin Lee",
  "category": "research",
  "topics": "safety-alignment",
  "orgs": null,
  "regions": null,
  "published_at": "2026-05-26T17:57:04.000Z",
  "fetched_at": "2026-07-14T16:30:23.248Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/3654",
  "original_url": "https://arxiv.org/abs/2605.27355v2",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}