{"as_of":"2026-08-21T23:12:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:4971ce413214dc68a3f70efe332ca3ec4c2926ddb0d3ad62673b5e38536b33a8","coverage":[{"denominator":4,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":4,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-06T22:52:33.805092Z","state":"measured"},{"denominator":4,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":4,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-21T06:32:19.484+00:00","state":"measured"},{"denominator":0,"lane":"inbound_itemization","note":"Pith citing papers itemized under the disclosed page cap.","records_observed":0,"source":"paper_references, paper_reference_links","source_observed_at":null,"state":"measured"},{"denominator":1,"lane":"external_citation_measurements","note":"A source-named dated measurement, never combined with another source.","records_observed":0,"source":"cited_works","source_observed_at":null,"state":"measured"}],"external_citation_measurements":[],"inbound":[],"links":{"evidence":"/evidence","html":"/paper/2506.20471/citation-record","integrity":"/paper/2506.20471/integrity","json":"/paper/2506.20471/citation-record.json","paper":"/paper/2506.20471"},"outbound":[{"citation":{"cited_paper":{"arxiv_id":"1909.08593","last_updated":"2020-01-08T23:02:36Z","snapshot_observed_at":"2026-08-21T16:54:14.450108Z","submitted_at":"2019-09-18T17:33:39Z","title":"Fine-Tuning Language Models from Human Preferences","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1909.08593","snapshot_observed_at":"2026-08-06T22:52:33.805092Z","title":"In proceedings of the ACM Conference on Health, Inference, and Learning, pages 110–120","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2506.20471","last_updated":"2025-06-25T14:19:57Z","snapshot_observed_at":"2026-08-18T18:59:34.898264Z","submitted_at":"2025-06-25T14:19:57Z","title":"Probing AI Safety with Source Code","version":1},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-08-06T22:52:33.805092Z"},"links":{"cited_paper":"/paper/1909.08593","citing_paper":"/paper/2506.20471"},"observation_digest":"sha256:a832aad41167345cf57fef73eb8ede22fe4cafc6a3b25d6a8b6e5f0beb086e45","observation_id":"2dee118d-3ce8-494a-9682-5ea3b603bfaf","resolution":{"observed_at":"2026-08-06T22:52:33.805092Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2004.09456","last_updated":"2020-04-20T17:14:33Z","snapshot_observed_at":"2026-08-18T18:59:00.636649Z","submitted_at":"2020-04-20T17:14:33Z","title":"StereoSet: Measuring stereotypical bias in pretrained language models","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2004.09456","snapshot_observed_at":"2026-08-06T22:52:33.800578Z","title":"arXiv preprint arXiv:2004.09456","venue":null,"work_id":null,"year":2004},"citing_paper":{"arxiv_id":"2506.20471","last_updated":"2025-06-25T14:19:57Z","snapshot_observed_at":"2026-08-18T18:59:34.898264Z","submitted_at":"2025-06-25T14:19:57Z","title":"Probing AI Safety with Source Code","version":1},"reference_index":2020,"source":"pdf_text","source_observed_at":"2026-08-06T22:52:33.800578Z"},"links":{"cited_paper":"/paper/2004.09456","citing_paper":"/paper/2506.20471"},"observation_digest":"sha256:891a5cbda182e96e25a4a34a32b5ed9e6aa5c13978c0f70fd3e7cd7c214fed2a","observation_id":"8fc4d1c2-04b1-4cb5-8827-6806b1a5db60","resolution":{"observed_at":"2026-08-06T22:52:33.800578Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2303.08774","last_updated":"2024-03-04T06:01:33Z","snapshot_observed_at":"2026-08-17T09:58:46.058102Z","submitted_at":"2023-03-15T17:15:04Z","title":"GPT-4 Technical Report","version":6},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2303.08774","snapshot_observed_at":"2026-08-06T22:52:33.790582Z","title":"Nature Machine Intelligence , 3(6):461–463","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2506.20471","last_updated":"2025-06-25T14:19:57Z","snapshot_observed_at":"2026-08-18T18:59:34.898264Z","submitted_at":"2025-06-25T14:19:57Z","title":"Probing AI Safety with Source Code","version":1},"reference_index":2021,"source":"pdf_text","source_observed_at":"2026-08-06T22:52:33.790582Z"},"links":{"cited_paper":"/paper/2303.08774","citing_paper":"/paper/2506.20471"},"observation_digest":"sha256:bf08defaebd2197466375d7845db4b0127f846a5bfafafe8f8f3daf13253ce30","observation_id":"9cba7311-bda5-4d6f-b1e8-e0453a1f6fdc","resolution":{"observed_at":"2026-08-06T22:52:33.790582Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2009.11462","last_updated":"2020-09-25T20:22:26Z","snapshot_observed_at":"2026-08-18T05:13:24.803705Z","submitted_at":"2020-09-24T03:17:19Z","title":"RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2009.11462","snapshot_observed_at":"2026-08-06T22:52:33.795772Z","title":"Applied Intelligence, 53(7):8421–8435","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2506.20471","last_updated":"2025-06-25T14:19:57Z","snapshot_observed_at":"2026-08-18T18:59:34.898264Z","submitted_at":"2025-06-25T14:19:57Z","title":"Probing AI Safety with Source Code","version":1},"reference_index":2023,"source":"pdf_text","source_observed_at":"2026-08-06T22:52:33.795772Z"},"links":{"cited_paper":"/paper/2009.11462","citing_paper":"/paper/2506.20471"},"observation_digest":"sha256:c75ce6f3d48cfcfaa4aec2c9288595a2a45188379e9d4108d20b7ca60ffd6519","observation_id":"008df72d-9ed6-4bc0-bf0e-59aa3955ca21","resolution":{"observed_at":"2026-08-06T22:52:33.795772Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}}],"paper":{"arxiv_id":"2506.20471","last_updated":"2025-06-25T14:19:57Z","latest_version":1,"primary_category":"cs.CL","snapshot_observed_at":"2026-08-18T18:59:34.898264Z","submitted_at":"2025-06-25T14:19:57Z","title":"Probing AI Safety with Source Code"},"reference_resolution":{"displayed":4,"state_counts":{"malformed_identifier":0,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":4,"verified_exact":0,"verified_fuzzy":0},"total_outbound_references":4},"refusal":"A citation records a reference. It does not transfer a finding from one paper to another.","schema":"pith.paper-citation-record.v1","standing_sources":[{"observed_at":"2026-08-21T06:32:19.484+00:00","source":"crossref"},{"observed_at":"2026-08-21T06:32:16.066871+00:00","source":"retraction_watch"}],"thesis":"As of 21 August 2026, this Paper Citation Record lists 4 of 4 outbound references and 0 inbound Pith citation observations for arXiv:2506.20471."}