{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:KXQ66VKFTUMYMDLD7UOUWWOZHB","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":"35f7622dbc102e411c9e4d0fbbf4ef0e9cf9934e654ffea5ad0935605a1fe062","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-06-06T04:04:57Z","title_canon_sha256":"fb15fad6ae76d0fbf74104687a192574e62487acf670be0527275f0d394517c1"},"schema_version":"1.0","source":{"id":"2606.07970","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.07970","created_at":"2026-06-09T01:04:56Z"},{"alias_kind":"arxiv_version","alias_value":"2606.07970v1","created_at":"2026-06-09T01:04:56Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.07970","created_at":"2026-06-09T01:04:56Z"},{"alias_kind":"pith_short_12","alias_value":"KXQ66VKFTUMY","created_at":"2026-06-09T01:04:56Z"},{"alias_kind":"pith_short_16","alias_value":"KXQ66VKFTUMYMDLD","created_at":"2026-06-09T01:04:56Z"},{"alias_kind":"pith_short_8","alias_value":"KXQ66VKF","created_at":"2026-06-09T01:04:56Z"}],"graph_snapshots":[{"event_id":"sha256:01c095fbf0c7fbcb02593f06d366431971a54559c9db070952cd7a7302d1ac33","target":"graph","created_at":"2026-06-09T01:04:56Z","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/2606.07970/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Current open-weight large language models (LLMs) are prone to malicious finetuning attacks, which could compromise the safety alignment of LLMs with only a few steps of supervised finetuning (SFT) on poisoned datasets. Existing alignment-stage defenses are primarily designed to defend against attacks that use parameter-efficient finetuning methods. However, they fail to defend against stronger attacks that use full-parameter finetuning. In this paper, we propose Patcher, a method inspired by adversarial training and bi-level optimization, to combat such attacks. Patcher strengthens the simulat","authors_text":"Haoming Wen, Jingzhao Zhang, Minrui Luo, Qingyu Shi, Shi Chen, Siyuan Liu, Tianxing He","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-06-06T04:04:57Z","title":"Defending Against Malicious Finetuning by Scaling Train-time Adversarial Attacks"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.07970","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:d05ea14deedac1e94fc06781d6bfdd204037ce2a76a6fc17ebc48be17e9857dd","target":"record","created_at":"2026-06-09T01:04:56Z","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":"35f7622dbc102e411c9e4d0fbbf4ef0e9cf9934e654ffea5ad0935605a1fe062","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-06-06T04:04:57Z","title_canon_sha256":"fb15fad6ae76d0fbf74104687a192574e62487acf670be0527275f0d394517c1"},"schema_version":"1.0","source":{"id":"2606.07970","kind":"arxiv","version":1}},"canonical_sha256":"55e1ef55459d19860d63fd1d4b59d93855214527de41a5d92176fec079980725","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"55e1ef55459d19860d63fd1d4b59d93855214527de41a5d92176fec079980725","first_computed_at":"2026-06-09T01:04:56.805075Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-09T01:04:56.805075Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"RcDAzp0fIy+/GRWi8waQ/5e1XA7aGt7GGW2+t0r60xMPFUKpGLPCO5/BkbrF4Y44+1BFBbgkeSSpqLiaMeiyAw==","signature_status":"signed_v1","signed_at":"2026-06-09T01:04:56.807709Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.07970","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:d05ea14deedac1e94fc06781d6bfdd204037ce2a76a6fc17ebc48be17e9857dd","sha256:01c095fbf0c7fbcb02593f06d366431971a54559c9db070952cd7a7302d1ac33"],"state_sha256":"9fd1dda135a45e65bbeae363c4586aa912b15c7ba8d7f733ab168c6a5c135548"}