{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:5IF44TUC5PZL4JV2KQRNSBZQ6Q","short_pith_number":"pith:5IF44TUC","schema_version":"1.0","canonical_sha256":"ea0bce4e82ebf2be26ba5422d90730f434b2ddab499699f5465bccb2c29e66f6","source":{"kind":"arxiv","id":"2404.14068","version":1},"attestation_state":"computed","paper":{"title":"Holistic Safety and Responsibility Evaluations of Advanced AI Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Allan Dafoe, Christina Butterfield, Dawn Bloxwich, Iason Gabriel, Jennifer Beroshi, Jenny Brennan, Jilin Chen, Joslyn Barnhart, Laura Weidinger, Lev Proleev, Lewis Ho, Lisa Anne Hendricks, Mikel Rodriguez, Oscar Chang, Ramona Comanescu, Sebastian Farquhar, Susie Young, Will Hawkins, William Isaac","submitted_at":"2024-04-22T10:26:49Z","abstract_excerpt":"Safety and responsibility evaluations of advanced AI models are a critical but developing field of research and practice. In the development of Google DeepMind's advanced AI models, we innovated on and applied a broad set of approaches to safety evaluation. In this report, we summarise and share elements of our evolving approach as well as lessons learned for a broad audience. Key lessons learned include: First, theoretical underpinnings and frameworks are invaluable to organise the breadth of risk domains, modalities, forms, metrics, and goals. Second, theory and practice of safety evaluation"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2404.14068","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2024-04-22T10:26:49Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"d98bf5dce0759bf1bb86571d2ae64d3210a6efa45361e5419061309d5c0c16d5","abstract_canon_sha256":"6c4c4a9eac9dd2955662cdeb148d7a376278643b9c8a28ed1a85497b5c0c4a57"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:10:41.193516Z","signature_b64":"uCA6GroqYZ4v08A6vqslMXNuRxs/7AgHq66v6uakHgRYz46n2yfNL96I3hUgCmfv9Jvws5QGXWMzzMakyPimCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ea0bce4e82ebf2be26ba5422d90730f434b2ddab499699f5465bccb2c29e66f6","last_reissued_at":"2026-07-05T08:10:41.193062Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:10:41.193062Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Holistic Safety and Responsibility Evaluations of Advanced AI Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Allan Dafoe, Christina Butterfield, Dawn Bloxwich, Iason Gabriel, Jennifer Beroshi, Jenny Brennan, Jilin Chen, Joslyn Barnhart, Laura Weidinger, Lev Proleev, Lewis Ho, Lisa Anne Hendricks, Mikel Rodriguez, Oscar Chang, Ramona Comanescu, Sebastian Farquhar, Susie Young, Will Hawkins, William Isaac","submitted_at":"2024-04-22T10:26:49Z","abstract_excerpt":"Safety and responsibility evaluations of advanced AI models are a critical but developing field of research and practice. In the development of Google DeepMind's advanced AI models, we innovated on and applied a broad set of approaches to safety evaluation. In this report, we summarise and share elements of our evolving approach as well as lessons learned for a broad audience. Key lessons learned include: First, theoretical underpinnings and frameworks are invaluable to organise the breadth of risk domains, modalities, forms, metrics, and goals. Second, theory and practice of safety evaluation"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.14068","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2404.14068/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2404.14068","created_at":"2026-07-05T08:10:41.193119+00:00"},{"alias_kind":"arxiv_version","alias_value":"2404.14068v1","created_at":"2026-07-05T08:10:41.193119+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.14068","created_at":"2026-07-05T08:10:41.193119+00:00"},{"alias_kind":"pith_short_12","alias_value":"5IF44TUC5PZL","created_at":"2026-07-05T08:10:41.193119+00:00"},{"alias_kind":"pith_short_16","alias_value":"5IF44TUC5PZL4JV2","created_at":"2026-07-05T08:10:41.193119+00:00"},{"alias_kind":"pith_short_8","alias_value":"5IF44TUC","created_at":"2026-07-05T08:10:41.193119+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.10154","citing_title":"Quality Is Not a Safety Proxy Under Quantization","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2507.06261","citing_title":"Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities","ref_index":88,"is_internal_anchor":false},{"citing_arxiv_id":"2508.19932","citing_title":"CASE: An Agentic AI Framework for Enhancing Scam Intelligence in Digital Payments","ref_index":4,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5IF44TUC5PZL4JV2KQRNSBZQ6Q","json":"https://pith.science/pith/5IF44TUC5PZL4JV2KQRNSBZQ6Q.json","graph_json":"https://pith.science/api/pith-number/5IF44TUC5PZL4JV2KQRNSBZQ6Q/graph.json","events_json":"https://pith.science/api/pith-number/5IF44TUC5PZL4JV2KQRNSBZQ6Q/events.json","paper":"https://pith.science/paper/5IF44TUC"},"agent_actions":{"view_html":"https://pith.science/pith/5IF44TUC5PZL4JV2KQRNSBZQ6Q","download_json":"https://pith.science/pith/5IF44TUC5PZL4JV2KQRNSBZQ6Q.json","view_paper":"https://pith.science/paper/5IF44TUC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2404.14068&json=true","fetch_graph":"https://pith.science/api/pith-number/5IF44TUC5PZL4JV2KQRNSBZQ6Q/graph.json","fetch_events":"https://pith.science/api/pith-number/5IF44TUC5PZL4JV2KQRNSBZQ6Q/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5IF44TUC5PZL4JV2KQRNSBZQ6Q/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5IF44TUC5PZL4JV2KQRNSBZQ6Q/action/storage_attestation","attest_author":"https://pith.science/pith/5IF44TUC5PZL4JV2KQRNSBZQ6Q/action/author_attestation","sign_citation":"https://pith.science/pith/5IF44TUC5PZL4JV2KQRNSBZQ6Q/action/citation_signature","submit_replication":"https://pith.science/pith/5IF44TUC5PZL4JV2KQRNSBZQ6Q/action/replication_record"}},"created_at":"2026-07-05T08:10:41.193119+00:00","updated_at":"2026-07-05T08:10:41.193119+00:00"}