{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:OWZLOWUR4RN22EUFGPMDPRSFMS","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":"84d4d599a91bb86e0fdb3a7b4420dae9f3b0daf4ae843a4fc11593615ae0e485","cross_cats_sorted":["cs.CL","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2025-06-05T17:59:55Z","title_canon_sha256":"2df6fd9183487b510367d6d197893ed5e7823df56e620c4a0242204d0fc94dfd"},"schema_version":"1.0","source":{"id":"2506.05346","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2506.05346","created_at":"2026-07-05T11:16:46Z"},{"alias_kind":"arxiv_version","alias_value":"2506.05346v1","created_at":"2026-07-05T11:16:46Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.05346","created_at":"2026-07-05T11:16:46Z"},{"alias_kind":"pith_short_12","alias_value":"OWZLOWUR4RN2","created_at":"2026-07-05T11:16:46Z"},{"alias_kind":"pith_short_16","alias_value":"OWZLOWUR4RN22EUF","created_at":"2026-07-05T11:16:46Z"},{"alias_kind":"pith_short_8","alias_value":"OWZLOWUR","created_at":"2026-07-05T11:16:46Z"}],"graph_snapshots":[{"event_id":"sha256:cfb68c78b24ed80238c9dbf8cc4239a627407472526b23aa2b03b2c47655ad5e","target":"graph","created_at":"2026-07-05T11:16:46Z","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/2506.05346/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Recent advancements in large language models (LLMs) have underscored their vulnerability to safety alignment jailbreaks, particularly when subjected to downstream fine-tuning. However, existing mitigation strategies primarily focus on reactively addressing jailbreak incidents after safety guardrails have been compromised, removing harmful gradients during fine-tuning, or continuously reinforcing safety alignment throughout fine-tuning. As such, they tend to overlook a critical upstream factor: the role of the original safety-alignment data. This paper therefore investigates the degradation of ","authors_text":"Lei Hsiung, Linyue Song, Pin-Yu Chen, Tianyu Pang, Tsung-Yi Ho, Yaoqing Yang, Yung-Chen Tang","cross_cats":["cs.CL","cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2025-06-05T17:59:55Z","title":"Why LLM Safety Guardrails Collapse After Fine-tuning: A Similarity Analysis Between Alignment and Fine-tuning Datasets"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.05346","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":"sha256:4ce5557afa0e56eed6de8695091b6bb9f7341cd12d0cf9de259d1ad7e92d5a85","target":"record","created_at":"2026-07-05T11:16:46Z","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":"84d4d599a91bb86e0fdb3a7b4420dae9f3b0daf4ae843a4fc11593615ae0e485","cross_cats_sorted":["cs.CL","cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2025-06-05T17:59:55Z","title_canon_sha256":"2df6fd9183487b510367d6d197893ed5e7823df56e620c4a0242204d0fc94dfd"},"schema_version":"1.0","source":{"id":"2506.05346","kind":"arxiv","version":1}},"canonical_sha256":"75b2b75a91e45bad128533d837c6456490332fa02c570739ce549bbe9a0a68aa","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"75b2b75a91e45bad128533d837c6456490332fa02c570739ce549bbe9a0a68aa","first_computed_at":"2026-07-05T11:16:46.213593Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:16:46.213593Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"9UMiR+fOnO0o0bi6zPU26l3QvwrJo5idJ1tDj//dBGqyFika3Jyr2IJZoZ3yKhs3vpJgNzp5Mon376rEKsNDCA==","signature_status":"signed_v1","signed_at":"2026-07-05T11:16:46.214084Z","signed_message":"canonical_sha256_bytes"},"source_id":"2506.05346","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:4ce5557afa0e56eed6de8695091b6bb9f7341cd12d0cf9de259d1ad7e92d5a85","sha256:cfb68c78b24ed80238c9dbf8cc4239a627407472526b23aa2b03b2c47655ad5e"],"state_sha256":"2d2cfa7a30f3a0020a501c463749ef617ec8de65993943bf85b8b6d46a6147da"}