{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:WNLHMCPU3H4JIZ6DE27B57DOUE","short_pith_number":"pith:WNLHMCPU","canonical_record":{"source":{"id":"2606.30899","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CR","submitted_at":"2026-06-29T20:40:21Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"b88f8edd953240022685755778c3610b15f7179f3281a43eeb90442cfbb02d97","abstract_canon_sha256":"74ee4a4804ef4e80206410d1dc786e83d4788f32a4962c8808e09c840da04a77"},"schema_version":"1.0"},"canonical_sha256":"b3567609f4d9f89467c326be1efc6ea13a8cd516d9c9ecef882362630ff25532","source":{"kind":"arxiv","id":"2606.30899","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.30899","created_at":"2026-07-01T00:17:20Z"},{"alias_kind":"arxiv_version","alias_value":"2606.30899v1","created_at":"2026-07-01T00:17:20Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.30899","created_at":"2026-07-01T00:17:20Z"},{"alias_kind":"pith_short_12","alias_value":"WNLHMCPU3H4J","created_at":"2026-07-01T00:17:20Z"},{"alias_kind":"pith_short_16","alias_value":"WNLHMCPU3H4JIZ6D","created_at":"2026-07-01T00:17:20Z"},{"alias_kind":"pith_short_8","alias_value":"WNLHMCPU","created_at":"2026-07-01T00:17:20Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:WNLHMCPU3H4JIZ6DE27B57DOUE","target":"record","payload":{"canonical_record":{"source":{"id":"2606.30899","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CR","submitted_at":"2026-06-29T20:40:21Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"b88f8edd953240022685755778c3610b15f7179f3281a43eeb90442cfbb02d97","abstract_canon_sha256":"74ee4a4804ef4e80206410d1dc786e83d4788f32a4962c8808e09c840da04a77"},"schema_version":"1.0"},"canonical_sha256":"b3567609f4d9f89467c326be1efc6ea13a8cd516d9c9ecef882362630ff25532","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-01T00:17:20.872902Z","signature_b64":"9aVKLPWhPBq0yXHIotkqnAQev+1/1mdKuaOfanOPglTvR3RzcUw9as3ALtPNKvGQ7uF4+DNxQgCrIqsRkTMABQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b3567609f4d9f89467c326be1efc6ea13a8cd516d9c9ecef882362630ff25532","last_reissued_at":"2026-07-01T00:17:20.872445Z","signature_status":"signed_v1","first_computed_at":"2026-07-01T00:17:20.872445Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2606.30899","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-01T00:17:20Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"h3bcfE25jITHuaEfhfE8C4ok8U4S0rITeHQ6htOSAsWUH+BHrs9YrRg/XWBF8BrzRXJbWTonHy6QSUxN9EXqCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-02T07:04:32.071218Z"},"content_sha256":"1e34008e702de019a48376b99ba77197defd6d373313fb7166980f98238d6e54","schema_version":"1.0","event_id":"sha256:1e34008e702de019a48376b99ba77197defd6d373313fb7166980f98238d6e54"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:WNLHMCPU3H4JIZ6DE27B57DOUE","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Curvature-Guided Module Localization for Low-Rank Detoxification of Backdoored Large Language Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Andrew Arash Mahyari, Arash Raftari, Mehrdad Mahdavi, Nathan Blackthorn","submitted_at":"2026-06-29T20:40:21Z","abstract_excerpt":"Backdoor attacks pose a serious threat to large language models (LLMs) by causing otherwise benign systems to produce attacker-specified malicious behavior when a hidden trigger is present. In this work, we study post hoc detoxification of backdoored LLMs in a practical setting where the defender has access to the poisoned model but does not wish to retrain the full network from scratch. We propose a mechanistically guided weight-space repair framework that first localizes modules involved in propagating trigger-induced behavior using activation patching and Fisher/K-FAC curvature analysis, an"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.30899","kind":"arxiv","version":1},"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/2606.30899/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":"2026-07-01T00:17:20Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"7agFCeRyfGiD7CJ47ReekBxG+aeq52hgJpTTSDeo5/N4Ow7uD10kvPRjIOuiwtbTGl6qvHG6Gvynmx5xjR3IDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-02T07:04:32.071618Z"},"content_sha256":"0e9c1d2f3197c086d48766afe60bdad1a6dc8a0692b650e6b67091f1927b4acf","schema_version":"1.0","event_id":"sha256:0e9c1d2f3197c086d48766afe60bdad1a6dc8a0692b650e6b67091f1927b4acf"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/WNLHMCPU3H4JIZ6DE27B57DOUE/bundle.json","state_url":"https://pith.science/pith/WNLHMCPU3H4JIZ6DE27B57DOUE/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/WNLHMCPU3H4JIZ6DE27B57DOUE/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":"2026-07-02T07:04:32Z","links":{"resolver":"https://pith.science/pith/WNLHMCPU3H4JIZ6DE27B57DOUE","bundle":"https://pith.science/pith/WNLHMCPU3H4JIZ6DE27B57DOUE/bundle.json","state":"https://pith.science/pith/WNLHMCPU3H4JIZ6DE27B57DOUE/state.json","well_known_bundle":"https://pith.science/.well-known/pith/WNLHMCPU3H4JIZ6DE27B57DOUE/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:WNLHMCPU3H4JIZ6DE27B57DOUE","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":"74ee4a4804ef4e80206410d1dc786e83d4788f32a4962c8808e09c840da04a77","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CR","submitted_at":"2026-06-29T20:40:21Z","title_canon_sha256":"b88f8edd953240022685755778c3610b15f7179f3281a43eeb90442cfbb02d97"},"schema_version":"1.0","source":{"id":"2606.30899","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.30899","created_at":"2026-07-01T00:17:20Z"},{"alias_kind":"arxiv_version","alias_value":"2606.30899v1","created_at":"2026-07-01T00:17:20Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.30899","created_at":"2026-07-01T00:17:20Z"},{"alias_kind":"pith_short_12","alias_value":"WNLHMCPU3H4J","created_at":"2026-07-01T00:17:20Z"},{"alias_kind":"pith_short_16","alias_value":"WNLHMCPU3H4JIZ6D","created_at":"2026-07-01T00:17:20Z"},{"alias_kind":"pith_short_8","alias_value":"WNLHMCPU","created_at":"2026-07-01T00:17:20Z"}],"graph_snapshots":[{"event_id":"sha256:0e9c1d2f3197c086d48766afe60bdad1a6dc8a0692b650e6b67091f1927b4acf","target":"graph","created_at":"2026-07-01T00:17:20Z","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/2606.30899/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Backdoor attacks pose a serious threat to large language models (LLMs) by causing otherwise benign systems to produce attacker-specified malicious behavior when a hidden trigger is present. In this work, we study post hoc detoxification of backdoored LLMs in a practical setting where the defender has access to the poisoned model but does not wish to retrain the full network from scratch. We propose a mechanistically guided weight-space repair framework that first localizes modules involved in propagating trigger-induced behavior using activation patching and Fisher/K-FAC curvature analysis, an","authors_text":"Andrew Arash Mahyari, Arash Raftari, Mehrdad Mahdavi, Nathan Blackthorn","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CR","submitted_at":"2026-06-29T20:40:21Z","title":"Curvature-Guided Module Localization for Low-Rank Detoxification of Backdoored Large Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.30899","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":"sha256:1e34008e702de019a48376b99ba77197defd6d373313fb7166980f98238d6e54","target":"record","created_at":"2026-07-01T00:17:20Z","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":"74ee4a4804ef4e80206410d1dc786e83d4788f32a4962c8808e09c840da04a77","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CR","submitted_at":"2026-06-29T20:40:21Z","title_canon_sha256":"b88f8edd953240022685755778c3610b15f7179f3281a43eeb90442cfbb02d97"},"schema_version":"1.0","source":{"id":"2606.30899","kind":"arxiv","version":1}},"canonical_sha256":"b3567609f4d9f89467c326be1efc6ea13a8cd516d9c9ecef882362630ff25532","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"b3567609f4d9f89467c326be1efc6ea13a8cd516d9c9ecef882362630ff25532","first_computed_at":"2026-07-01T00:17:20.872445Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-01T00:17:20.872445Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"9aVKLPWhPBq0yXHIotkqnAQev+1/1mdKuaOfanOPglTvR3RzcUw9as3ALtPNKvGQ7uF4+DNxQgCrIqsRkTMABQ==","signature_status":"signed_v1","signed_at":"2026-07-01T00:17:20.872902Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.30899","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:1e34008e702de019a48376b99ba77197defd6d373313fb7166980f98238d6e54","sha256:0e9c1d2f3197c086d48766afe60bdad1a6dc8a0692b650e6b67091f1927b4acf"],"state_sha256":"a901487bde6af8ebb238215b669421109f8f48a96cf90a4516cfe3d5761c905e"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Bbhiz1WuUaI1hr5pAb/3q+LwJlb9stFaPcjpElE90RoJdCktw+T3GM94nqiGrYPNwfg6z0fBwRzAT4DuhPD9Dg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-02T07:04:32.073724Z","bundle_sha256":"43610d4e9ecba77ac7a5efc4d3870db756a1cd3fcacd4f2a101c7663cfec257e"}}