{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:5MN5WHBGO7D6IB7VEYLZAYCP4E","short_pith_number":"pith:5MN5WHBG","canonical_record":{"source":{"id":"2506.08473","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-06-10T05:59:48Z","cross_cats_sorted":[],"title_canon_sha256":"3282fc0329ac8b465edd3ca9a6bb6c6c605208e464aa3f817440b571dd7a8921","abstract_canon_sha256":"752850bdb13c1d7a40831be7690391883c45824ad6916b6a1cc25b2add5edc3c"},"schema_version":"1.0"},"canonical_sha256":"eb1bdb1c2677c7e407f5261790604fe11f6ac923d0625d17d3dc6b3be19a0d30","source":{"kind":"arxiv","id":"2506.08473","version":4},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2506.08473","created_at":"2026-06-11T01:10:26Z"},{"alias_kind":"arxiv_version","alias_value":"2506.08473v4","created_at":"2026-06-11T01:10:26Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.08473","created_at":"2026-06-11T01:10:26Z"},{"alias_kind":"pith_short_12","alias_value":"5MN5WHBGO7D6","created_at":"2026-06-11T01:10:26Z"},{"alias_kind":"pith_short_16","alias_value":"5MN5WHBGO7D6IB7V","created_at":"2026-06-11T01:10:26Z"},{"alias_kind":"pith_short_8","alias_value":"5MN5WHBG","created_at":"2026-06-11T01:10:26Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:5MN5WHBGO7D6IB7VEYLZAYCP4E","target":"record","payload":{"canonical_record":{"source":{"id":"2506.08473","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-06-10T05:59:48Z","cross_cats_sorted":[],"title_canon_sha256":"3282fc0329ac8b465edd3ca9a6bb6c6c605208e464aa3f817440b571dd7a8921","abstract_canon_sha256":"752850bdb13c1d7a40831be7690391883c45824ad6916b6a1cc25b2add5edc3c"},"schema_version":"1.0"},"canonical_sha256":"eb1bdb1c2677c7e407f5261790604fe11f6ac923d0625d17d3dc6b3be19a0d30","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-11T01:10:26.770234Z","signature_b64":"uI4T+nRyl6whF6jcYHpuqgHt3M4jTos5L4LuUDvDnpbRP6Ysw/4UO/Q5dz6PrFFqNvrQ01XJhmBXNQYEPlqrCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"eb1bdb1c2677c7e407f5261790604fe11f6ac923d0625d17d3dc6b3be19a0d30","last_reissued_at":"2026-06-11T01:10:26.769367Z","signature_status":"signed_v1","first_computed_at":"2026-06-11T01:10:26.769367Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2506.08473","source_version":4,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-11T01:10:26Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"rVAALwrBCSAVxurHddW9RKNYDe6KXp1jwynPl7RtCaGnZ+RCcqZ22mOn0/Bxh+zWRe4hs8CZ37+Mi2tgvcn+Bg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-11T07:30:55.062563Z"},"content_sha256":"6dafe73681bb49544ed8c7ac87df5ae07cf3d1b7c8c91d82f2b4378c4c07ac88","schema_version":"1.0","event_id":"sha256:6dafe73681bb49544ed8c7ac87df5ae07cf3d1b7c8c91d82f2b4378c4c07ac88"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:5MN5WHBGO7D6IB7VEYLZAYCP4E","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"AsFT: Anchoring Safety During LLM Fine-Tuning Within Narrow Safety Basin","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Hailiang Dai, Jiayu Yao, Jigang Wang, Kunpeng Ning, Li Yuan, Qihui Zhang, Shuo Yang, Xiaojun Jia, Yibing Song, Yuyang Liu","submitted_at":"2025-06-10T05:59:48Z","abstract_excerpt":"Fine-tuning large language models (LLMs) improves performance but introduces critical safety vulnerabilities: even minimal harmful data can severely compromise safety measures. We observe that perturbations orthogonal to the alignment direction - defined by weight differences between aligned (safe) and unaligned models - rapidly compromise model safety. In contrast, updates along the alignment direction largely preserve it, revealing the parameter space as a \"narrow safety basin\". To address this, we propose AsFT (Anchoring Safety in Fine-Tuning) to maintain safety by explicitly constraining u"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.08473","kind":"arxiv","version":4},"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/2506.08473/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-11T01:10:26Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"MHaqQ8MM86+X0h68KmrspogFdnzgZ+YKj56P50o3L4gIXmUiI8M4yzM37AcCVnVAp64TtzgxynE99+qkrRq9DQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-11T07:30:55.063377Z"},"content_sha256":"aeb4dd96e45cc139ee02844e8e29be0a5d1d8ca78179cb949b2e7fa63f355b54","schema_version":"1.0","event_id":"sha256:aeb4dd96e45cc139ee02844e8e29be0a5d1d8ca78179cb949b2e7fa63f355b54"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/5MN5WHBGO7D6IB7VEYLZAYCP4E/bundle.json","state_url":"https://pith.science/pith/5MN5WHBGO7D6IB7VEYLZAYCP4E/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/5MN5WHBGO7D6IB7VEYLZAYCP4E/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-06-11T07:30:55Z","links":{"resolver":"https://pith.science/pith/5MN5WHBGO7D6IB7VEYLZAYCP4E","bundle":"https://pith.science/pith/5MN5WHBGO7D6IB7VEYLZAYCP4E/bundle.json","state":"https://pith.science/pith/5MN5WHBGO7D6IB7VEYLZAYCP4E/state.json","well_known_bundle":"https://pith.science/.well-known/pith/5MN5WHBGO7D6IB7VEYLZAYCP4E/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:5MN5WHBGO7D6IB7VEYLZAYCP4E","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":"752850bdb13c1d7a40831be7690391883c45824ad6916b6a1cc25b2add5edc3c","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-06-10T05:59:48Z","title_canon_sha256":"3282fc0329ac8b465edd3ca9a6bb6c6c605208e464aa3f817440b571dd7a8921"},"schema_version":"1.0","source":{"id":"2506.08473","kind":"arxiv","version":4}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2506.08473","created_at":"2026-06-11T01:10:26Z"},{"alias_kind":"arxiv_version","alias_value":"2506.08473v4","created_at":"2026-06-11T01:10:26Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.08473","created_at":"2026-06-11T01:10:26Z"},{"alias_kind":"pith_short_12","alias_value":"5MN5WHBGO7D6","created_at":"2026-06-11T01:10:26Z"},{"alias_kind":"pith_short_16","alias_value":"5MN5WHBGO7D6IB7V","created_at":"2026-06-11T01:10:26Z"},{"alias_kind":"pith_short_8","alias_value":"5MN5WHBG","created_at":"2026-06-11T01:10:26Z"}],"graph_snapshots":[{"event_id":"sha256:aeb4dd96e45cc139ee02844e8e29be0a5d1d8ca78179cb949b2e7fa63f355b54","target":"graph","created_at":"2026-06-11T01:10:26Z","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/2506.08473/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Fine-tuning large language models (LLMs) improves performance but introduces critical safety vulnerabilities: even minimal harmful data can severely compromise safety measures. We observe that perturbations orthogonal to the alignment direction - defined by weight differences between aligned (safe) and unaligned models - rapidly compromise model safety. In contrast, updates along the alignment direction largely preserve it, revealing the parameter space as a \"narrow safety basin\". To address this, we propose AsFT (Anchoring Safety in Fine-Tuning) to maintain safety by explicitly constraining u","authors_text":"Hailiang Dai, Jiayu Yao, Jigang Wang, Kunpeng Ning, Li Yuan, Qihui Zhang, Shuo Yang, Xiaojun Jia, Yibing Song, Yuyang Liu","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-06-10T05:59:48Z","title":"AsFT: Anchoring Safety During LLM Fine-Tuning Within Narrow Safety Basin"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.08473","kind":"arxiv","version":4},"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":"sha256:6dafe73681bb49544ed8c7ac87df5ae07cf3d1b7c8c91d82f2b4378c4c07ac88","target":"record","created_at":"2026-06-11T01:10:26Z","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":"752850bdb13c1d7a40831be7690391883c45824ad6916b6a1cc25b2add5edc3c","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-06-10T05:59:48Z","title_canon_sha256":"3282fc0329ac8b465edd3ca9a6bb6c6c605208e464aa3f817440b571dd7a8921"},"schema_version":"1.0","source":{"id":"2506.08473","kind":"arxiv","version":4}},"canonical_sha256":"eb1bdb1c2677c7e407f5261790604fe11f6ac923d0625d17d3dc6b3be19a0d30","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"eb1bdb1c2677c7e407f5261790604fe11f6ac923d0625d17d3dc6b3be19a0d30","first_computed_at":"2026-06-11T01:10:26.769367Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-11T01:10:26.769367Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"uI4T+nRyl6whF6jcYHpuqgHt3M4jTos5L4LuUDvDnpbRP6Ysw/4UO/Q5dz6PrFFqNvrQ01XJhmBXNQYEPlqrCw==","signature_status":"signed_v1","signed_at":"2026-06-11T01:10:26.770234Z","signed_message":"canonical_sha256_bytes"},"source_id":"2506.08473","source_kind":"arxiv","source_version":4}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:6dafe73681bb49544ed8c7ac87df5ae07cf3d1b7c8c91d82f2b4378c4c07ac88","sha256:aeb4dd96e45cc139ee02844e8e29be0a5d1d8ca78179cb949b2e7fa63f355b54"],"state_sha256":"9e59930f2e1bb9b091fd9ef9d7eb46ef7bc2153905894919962cb598bd05ee60"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"8x4sEEUalF7EuhpA8UCm/JZ/ocsKHKcemjDuQ1kTBbfh7uoWrw/0FA6/JkEl7henSk9yzCiJt7wndFuKnUzvDQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-11T07:30:55.067213Z","bundle_sha256":"2b98319a60abe04b9016deffc9bbf3d6b8f45522ab46f7efe56b608741467576"}}