{
  "id": 6932,
  "url": "https://arxiv.org/abs/2604.13076v3",
  "title": "Alignment midtraining for animals",
  "summary": "We investigate the robustness of value alignment via midtraining with synthetic documents, using animal compassion as a value that is both important in its own right and orthogonal to existing alignment efforts. To evaluate compassionate reasoning, we develop and publicly release Animal Norms In Moral Assessment (ANIMA), a 26-question evaluation spanning 13 ethical dimensions, publicly available as a dataset and Inspect evaluation. On ANIMA, training with 3000 documents achieves 77% compared to ",
  "authors": "Jasmine Brazilek, Miles Tidmarsh",
  "category": "research",
  "topics": "safety-alignment,finance-investment",
  "orgs": null,
  "regions": null,
  "published_at": "2026-03-21T01:32:24.000Z",
  "fetched_at": "2026-07-14T16:32:50.147Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/6932",
  "original_url": "https://arxiv.org/abs/2604.13076v3",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}