{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:6J3DFGYFRAFTOLFHIH7YFU7VCM","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":"b4bd2d51feb55f5ebaab40885af83e93db221c0d26d61ed1dcd29b5784fd1099","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-13T03:36:06Z","title_canon_sha256":"cb2d1a39f6dea5a33805a666a2cfbe8d047229da3e7d99665298cb19c9d5ec40"},"schema_version":"1.0","source":{"id":"2504.09420","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2504.09420","created_at":"2026-07-05T10:48:36Z"},{"alias_kind":"arxiv_version","alias_value":"2504.09420v1","created_at":"2026-07-05T10:48:36Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.09420","created_at":"2026-07-05T10:48:36Z"},{"alias_kind":"pith_short_12","alias_value":"6J3DFGYFRAFT","created_at":"2026-07-05T10:48:36Z"},{"alias_kind":"pith_short_16","alias_value":"6J3DFGYFRAFTOLFH","created_at":"2026-07-05T10:48:36Z"},{"alias_kind":"pith_short_8","alias_value":"6J3DFGYF","created_at":"2026-07-05T10:48:36Z"}],"graph_snapshots":[{"event_id":"sha256:e635ef2f6f4b4d266e2c776a478220b93b53641b07741fce790ef86df079b1a0","target":"graph","created_at":"2026-07-05T10:48:36Z","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/2504.09420/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Current safety alignment techniques for large language models (LLMs) face two key challenges: (1) under-generalization, which leaves models vulnerable to novel jailbreak attacks, and (2) over-alignment, which leads to the excessive refusal of benign instructions. Our preliminary investigation reveals semantic overlap between jailbreak/harmful queries and normal prompts in embedding space, suggesting that more effective safety alignment requires a deeper semantic understanding. This motivates us to incorporate safety-policy-driven reasoning into the alignment process. To this end, we propose th","authors_text":"Shikun Zhang, Wei Ye, Yutao Mou, Yuxiao Luo","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-13T03:36:06Z","title":"SaRO: Enhancing LLM Safety through Reasoning-based Alignment"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.09420","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:d6a40d2ac06391f8d3277def371466d95b8adda56a763d08d6d616cdffb899e5","target":"record","created_at":"2026-07-05T10:48:36Z","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":"b4bd2d51feb55f5ebaab40885af83e93db221c0d26d61ed1dcd29b5784fd1099","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-13T03:36:06Z","title_canon_sha256":"cb2d1a39f6dea5a33805a666a2cfbe8d047229da3e7d99665298cb19c9d5ec40"},"schema_version":"1.0","source":{"id":"2504.09420","kind":"arxiv","version":1}},"canonical_sha256":"f276329b05880b372ca741ff82d3f5130b7a79059eff5ab0213114ede0e85ed0","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"f276329b05880b372ca741ff82d3f5130b7a79059eff5ab0213114ede0e85ed0","first_computed_at":"2026-07-05T10:48:36.832872Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:48:36.832872Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"9cqSCK1L1GUf9/n3Dq5Dtb3tT4ptvhEi5Pvo7eWiFYP2HCvN6HcPkoad9XxLz5zUrgiSjOu0d7StHvv8idSJBQ==","signature_status":"signed_v1","signed_at":"2026-07-05T10:48:36.833324Z","signed_message":"canonical_sha256_bytes"},"source_id":"2504.09420","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:d6a40d2ac06391f8d3277def371466d95b8adda56a763d08d6d616cdffb899e5","sha256:e635ef2f6f4b4d266e2c776a478220b93b53641b07741fce790ef86df079b1a0"],"state_sha256":"7c75a69907a7bc936026cb96eb17c5e8fffdbf681c7d98ca5d56860ac3481975"}