{
  "id": 1307,
  "url": "https://arxiv.org/abs/2606.08451v1",
  "title": "Sycophancy as a Multilingual Alignment Failure: How Safety Degrades Across Languages, Topics, and Models",
  "summary": "Safety-aligned large language models often exhibit sycophancy, which is the tendency to affirm users' opinions regardless of factual accuracy. Although well-studied in English, its manifestation in other languages remains largely unexamined, leaving billions of non-English speakers potentially vulnerable to model-validated misinformation. We present the first large-scale, multi-model evaluation of cross-lingual sycophancy, benchmarking \\textbf{six instruction-tuned models} across \\textbf{1.1 mil",
  "authors": "Arya Shah, Himanshu Beniwal, Mayank Singh, Chaklam Silpasuwanchai",
  "category": "research",
  "topics": "safety-alignment,misinformation",
  "orgs": null,
  "regions": null,
  "published_at": "2026-06-07T04:50:40.000Z",
  "fetched_at": "2026-07-14T14:15:12.455Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/1307",
  "original_url": "https://arxiv.org/abs/2606.08451v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}