{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:3ZIXJ6FTQYI6DYETH3DP4B2DRS","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"4cad5e89fb512dfa9556c0b89751fb3b89db59d02f8e795ff4ed3169dcd8277e","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-06-17T15:11:43Z","title_canon_sha256":"f30d3419cca7d7e939eec92cb171fed63ba54bf2662c466369d90b47df9b7f32"},"schema_version":"1.0","source":{"id":"2606.19168","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.19168","created_at":"2026-06-19T16:11:58Z"},{"alias_kind":"arxiv_version","alias_value":"2606.19168v1","created_at":"2026-06-19T16:11:58Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.19168","created_at":"2026-06-19T16:11:58Z"},{"alias_kind":"pith_short_12","alias_value":"3ZIXJ6FTQYI6","created_at":"2026-06-19T16:11:58Z"},{"alias_kind":"pith_short_16","alias_value":"3ZIXJ6FTQYI6DYET","created_at":"2026-06-19T16:11:58Z"},{"alias_kind":"pith_short_8","alias_value":"3ZIXJ6FT","created_at":"2026-06-19T16:11:58Z"}],"graph_snapshots":[{"event_id":"sha256:10099b7b2bdfe64ccb20d8c1b41b30f1efc9a978a1414509622e7bdc5deb887b","target":"graph","created_at":"2026-06-19T16:11:58Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2606.19168/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"To achieve deeper safety alignment for large language models (LLMs), recent efforts have studied how to push safety interventions earlier into the pretraining stage, primarily by filtering unsafe data or rewriting it into safer forms. We argue that pretraining-stage alignment should go beyond making the data safe: LLMs may compose seemingly benign knowledge and capabilities into unsafe behaviors. To this end, we propose Safety Reflection Pretraining, a pretraining-stage alignment method which regularly inserts short safety reflections into pretraining corpora to integrate self-monitoring direc","authors_text":"Jinhan Li, Kaifeng Lyu, Kexian Tang, Yihan Xu, Zhuorui Ye","cross_cats":["cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-06-17T15:11:43Z","title":"Beyond Safe Data: Pretraining-Stage Alignment with Regular Safety Reflection"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.19168","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:533e09ead6f8af5c6dd19d9e7d8752945d7b009805dc806f110b93e6901c229a","target":"record","created_at":"2026-06-19T16:11:58Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"4cad5e89fb512dfa9556c0b89751fb3b89db59d02f8e795ff4ed3169dcd8277e","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-06-17T15:11:43Z","title_canon_sha256":"f30d3419cca7d7e939eec92cb171fed63ba54bf2662c466369d90b47df9b7f32"},"schema_version":"1.0","source":{"id":"2606.19168","kind":"arxiv","version":1}},"canonical_sha256":"de5174f8b38611e1e0933ec6fe07438ca305c57ef9abc1ac802b900ab1aef6a9","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"de5174f8b38611e1e0933ec6fe07438ca305c57ef9abc1ac802b900ab1aef6a9","first_computed_at":"2026-06-19T16:11:58.424304Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-19T16:11:58.424304Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"O0obpV3pg2L+QiDMJNDwGHZYg5xY9wr8nCvsBEB5oz5V69miZdg6DgVJzYnBfYc0v9xh3iFirB9fD7gTicBlCQ==","signature_status":"signed_v1","signed_at":"2026-06-19T16:11:58.424676Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.19168","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:533e09ead6f8af5c6dd19d9e7d8752945d7b009805dc806f110b93e6901c229a","sha256:10099b7b2bdfe64ccb20d8c1b41b30f1efc9a978a1414509622e7bdc5deb887b"],"state_sha256":"f4ce94909cdd85eb6432e92f1fb715c0b6122308ac258bb6e7d74a6c72cd1b22"}