{
  "id": 6337,
  "url": "https://arxiv.org/abs/2604.04261v1",
  "title": "APPA: Adaptive Preference Pluralistic Alignment for Fair Federated RLHF of LLMs",
  "summary": "Aligning large language models (LLMs) with diverse human preferences requires pluralistic alignment, where a single model must respect the values of multiple distinct groups simultaneously. In federated reinforcement learning from human feedback (FedRLHF), these groups align a shared policy without centralizing preference data, which makes fair reward aggregation essential. Existing aggregation methods exhibit clear trade offs: average based aggregation systematically under aligns worst performi",
  "authors": "Mahmoud Srewa, Tianyu Zhao, Salma Elmalaki",
  "category": "research",
  "topics": "regulation,safety-alignment",
  "orgs": null,
  "regions": null,
  "published_at": "2026-04-05T20:48:36.000Z",
  "fetched_at": "2026-07-14T16:32:24.292Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/6337",
  "original_url": "https://arxiv.org/abs/2604.04261v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}