{
  "url": "https://appliedfrontier.nl/en/research/signal-9cf289b414fdf955/",
  "title": "Deliberative alignment: reasoning enables safer language models",
  "published": "2024-12-20",
  "excerpt": "Deliberative alignment: reasoning enables safer language models Introducing our new alignment strategy for o1 models, which are directly taught safety specifications and how to reason",
  "content_sha256": "38c4ff60c5a93b9f315e835b3830875dec9fccdfdc151fe568e05b8ff23151af",
  "extracted_chars": 194,
  "artifact_links": [],
  "extraction_scope": "official_rss_metadata",
  "full_text_available": false,
  "id": "signal-9cf289b414fdf955",
  "org": "OpenAI",
  "topic": "Mathematical reasoning",
  "source_kind": "provider_announcement",
  "status": "unreviewed-source-report",
  "analysis_method": "deterministic_metadata_triage_v1",
  "novelty_verified": false,
  "proof_checked": false,
  "peer_review_verified": false,
  "matched_terms": [
    "reasoning"
  ],
  "checked_at": "2026-10-29T22:08:07+00:00",
  "first_seen": "2026-10-29T22:08:07+00:00",
  "source_hash": "a2e8e459debd0de6f04213b29a62604101ae448b528437edc0a9dfd6101fd857",
  "source_url": "https://openai.com/index/deliberative-alignment",
  "language": "en",
  "indexing": "noindex_until_editorially_enriched"
}