{
  "id": 7365,
  "url": "https://arxiv.org/abs/2603.11253v2",
  "title": "LLMs Can Infer Political Alignment from Online Conversations",
  "summary": "Due to the correlational structure in our traits such as identities, cultures, and political attitudes, seemingly innocuous preferences like following a band or using a specific slang can reveal private traits. This possibility, especially when combined with massive, public social data and advanced computational methods, poses a fundamental privacy risk. As our data exposure online and the rapid advancement of AI are increasing the risk of misuse, it is critical to understand the capacity of lar",
  "authors": "Byunghwee Lee, Sangyeon Kim, Filippo Menczer, Yong-Yeol Ahn, Haewoon Kwak, Jisun An",
  "category": "research",
  "topics": "safety-alignment,privacy-surveillance",
  "orgs": null,
  "regions": null,
  "published_at": "2026-03-11T19:26:04.000Z",
  "fetched_at": "2026-07-14T16:33:12.388Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/7365",
  "original_url": "https://arxiv.org/abs/2603.11253v2",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}