{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:P3SQX66CK2YA3FQP2NTY7LANMP","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":"c4253add02d69d88900e8ee5aa3577cb35c6d6dd60e93625a1d54aa1a6240f4b","cross_cats_sorted":["cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-02-02T02:56:50Z","title_canon_sha256":"d69788c3c0f9bda340ffcf49b1ddd73b31f922ecf67220ae64a42c7229d4bcbb"},"schema_version":"1.0","source":{"id":"2402.01109","kind":"arxiv","version":6}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2402.01109","created_at":"2026-07-05T09:39:52Z"},{"alias_kind":"arxiv_version","alias_value":"2402.01109v6","created_at":"2026-07-05T09:39:52Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.01109","created_at":"2026-07-05T09:39:52Z"},{"alias_kind":"pith_short_12","alias_value":"P3SQX66CK2YA","created_at":"2026-07-05T09:39:52Z"},{"alias_kind":"pith_short_16","alias_value":"P3SQX66CK2YA3FQP","created_at":"2026-07-05T09:39:52Z"},{"alias_kind":"pith_short_8","alias_value":"P3SQX66C","created_at":"2026-07-05T09:39:52Z"}],"graph_snapshots":[{"event_id":"sha256:5bcdb146fc456882e8e2d56085d1e816c3329f6c3c712db69979585247e221a3","target":"graph","created_at":"2026-07-05T09:39:52Z","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/2402.01109/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"The new paradigm of finetuning-as-a-service introduces a new attack surface for Large Language Models (LLMs): a few harmful data uploaded by users can easily trick the finetuning to produce an alignment-broken model. We conduct an empirical analysis and uncover a \\textit{harmful embedding drift} phenomenon, showing a probable cause of the alignment-broken effect. Inspired by our findings, we propose Vaccine, a perturbation-aware alignment technique to mitigate the security risk of users finetuning. The core idea of Vaccine is to produce invariant hidden embeddings by progressively adding craft","authors_text":"Ling Liu, Sihao Hu, Tiansheng Huang","cross_cats":["cs.CL"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-02-02T02:56:50Z","title":"Vaccine: Perturbation-aware Alignment for Large Language Models against Harmful Fine-tuning Attack"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.01109","kind":"arxiv","version":6},"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:bbaaa27258bfb0c157a59be43dd0c963aec772beb878744ddc445163186574f5","target":"record","created_at":"2026-07-05T09:39:52Z","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":"c4253add02d69d88900e8ee5aa3577cb35c6d6dd60e93625a1d54aa1a6240f4b","cross_cats_sorted":["cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-02-02T02:56:50Z","title_canon_sha256":"d69788c3c0f9bda340ffcf49b1ddd73b31f922ecf67220ae64a42c7229d4bcbb"},"schema_version":"1.0","source":{"id":"2402.01109","kind":"arxiv","version":6}},"canonical_sha256":"7ee50bfbc256b00d960fd3678fac0d63cf146145c93bd5dc70b42f46fb2622fd","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"7ee50bfbc256b00d960fd3678fac0d63cf146145c93bd5dc70b42f46fb2622fd","first_computed_at":"2026-07-05T09:39:52.925409Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:39:52.925409Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"kJ9L9e6GGJwFffsC1hXHHXRioCJMwHX/7ACtB+6a4WhaYIdU3CZ6xw3xgPnMIMFw2hWBO+d5wFEzlBoRQbceBA==","signature_status":"signed_v1","signed_at":"2026-07-05T09:39:52.925899Z","signed_message":"canonical_sha256_bytes"},"source_id":"2402.01109","source_kind":"arxiv","source_version":6}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:bbaaa27258bfb0c157a59be43dd0c963aec772beb878744ddc445163186574f5","sha256:5bcdb146fc456882e8e2d56085d1e816c3329f6c3c712db69979585247e221a3"],"state_sha256":"31cf310bb9b7b18d1a6b5cc532510d54b980e2efc7a2a554d35aed5694140c1c"}