{
  "id": 4467,
  "url": "https://arxiv.org/abs/2605.15217v1",
  "title": "Fair outputs, Biased Internals: Causal Potency and Asymmetry of Latent Bias in LLMs for High-Stakes Decisions",
  "summary": "Instruction-tuned language models exhibit behavioural fairness in high-stakes decisions while retaining biased associations in their internal representations. However, whether these suppressed representations can affect model outputs - and whether such causal potency is symmetric across demographic groups - remains unknown. We investigate the use of open-weight models for mortgage underwriting using matched applications that differ only in racially-associated names and reveal a critical disconne",
  "authors": "Jagdish Tripathy, Marcus Buckmann",
  "category": "research",
  "topics": "bias-fairness,finance-investment",
  "orgs": null,
  "regions": null,
  "published_at": "2026-05-12T12:14:58.000Z",
  "fetched_at": "2026-07-14T16:31:03.576Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/4467",
  "original_url": "https://arxiv.org/abs/2605.15217v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}