{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:3PGRP2FFOB44BDHLLPXEOLZTNW","short_pith_number":"pith:3PGRP2FF","canonical_record":{"source":{"id":"2411.12701","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2024-11-19T18:11:36Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"6acb00bfb4a1d42a7c4b20b54b5bb87016830fab6d252c5403d387423ed1b647","abstract_canon_sha256":"794c44ceb8e6590dcec6d2363c20ca67d1a9e373fb44c61f6dab1872dbb423ae"},"schema_version":"1.0"},"canonical_sha256":"dbcd17e8a57079c08ceb5bee472f336d84f335d530befa3979ec4958ee0ed1f3","source":{"kind":"arxiv","id":"2411.12701","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2411.12701","created_at":"2026-07-05T10:14:48Z"},{"alias_kind":"arxiv_version","alias_value":"2411.12701v3","created_at":"2026-07-05T10:14:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.12701","created_at":"2026-07-05T10:14:48Z"},{"alias_kind":"pith_short_12","alias_value":"3PGRP2FFOB44","created_at":"2026-07-05T10:14:48Z"},{"alias_kind":"pith_short_16","alias_value":"3PGRP2FFOB44BDHL","created_at":"2026-07-05T10:14:48Z"},{"alias_kind":"pith_short_8","alias_value":"3PGRP2FF","created_at":"2026-07-05T10:14:48Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:3PGRP2FFOB44BDHLLPXEOLZTNW","target":"record","payload":{"canonical_record":{"source":{"id":"2411.12701","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2024-11-19T18:11:36Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"6acb00bfb4a1d42a7c4b20b54b5bb87016830fab6d252c5403d387423ed1b647","abstract_canon_sha256":"794c44ceb8e6590dcec6d2363c20ca67d1a9e373fb44c61f6dab1872dbb423ae"},"schema_version":"1.0"},"canonical_sha256":"dbcd17e8a57079c08ceb5bee472f336d84f335d530befa3979ec4958ee0ed1f3","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:14:48.147366Z","signature_b64":"9L8CjndVbA/DBUtL9rh1Cvbe8KWdO4qjsD6ACXjD827+QI7YiNdynLPkIbnUp14Bm0T4+20KPGZwLgZK+ucdBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"dbcd17e8a57079c08ceb5bee472f336d84f335d530befa3979ec4958ee0ed1f3","last_reissued_at":"2026-07-05T10:14:48.146895Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:14:48.146895Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2411.12701","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:14:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"q028cggyPL25CGeP7phs8BiMN/NQCIr40K3BExV8gMtM/A0eqCRULYYal2PdtgdSPMUQV0IWQYDTqTqdAR9vAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-13T07:12:40.745115Z"},"content_sha256":"447cf7130aa8d5a234ca754875e0074efed1256dbf28f26cdec981c8482f0f1c","schema_version":"1.0","event_id":"sha256:447cf7130aa8d5a234ca754875e0074efed1256dbf28f26cdec981c8482f0f1c"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:3PGRP2FFOB44BDHLLPXEOLZTNW","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"When Backdoors Speak: Understanding LLM Backdoor Attacks Through Model-Generated Explanations","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Huaizhi Ge, Qifan Wang, Ruixiang Tang, Yiming Li, Yongfeng Zhang","submitted_at":"2024-11-19T18:11:36Z","abstract_excerpt":"Large Language Models (LLMs) are known to be vulnerable to backdoor attacks, where triggers embedded in poisoned samples can maliciously alter LLMs' behaviors. In this paper, we move beyond attacking LLMs and instead examine backdoor attacks through the novel lens of natural language explanations. Specifically, we leverage LLMs' generative capabilities to produce human-readable explanations for their decisions, enabling direct comparisons between explanations for clean and poisoned samples. Our results show that backdoored models produce coherent explanations for clean inputs but diverse and l"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.12701","kind":"arxiv","version":3},"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/2411.12701/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":"2026-07-05T10:14:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"lr/73Sz0oU6sWg/gM6ogGK3Iyqut+2JajAwKZwF88/t3L72vAlSp662lj4q0sDssMcd40yBbjVOscYB6O2bXAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-13T07:12:40.745658Z"},"content_sha256":"c96479f7a2527048b8066075fa7f3ead71f414b7c3d947f8152240b3282ec781","schema_version":"1.0","event_id":"sha256:c96479f7a2527048b8066075fa7f3ead71f414b7c3d947f8152240b3282ec781"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/3PGRP2FFOB44BDHLLPXEOLZTNW/bundle.json","state_url":"https://pith.science/pith/3PGRP2FFOB44BDHLLPXEOLZTNW/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/3PGRP2FFOB44BDHLLPXEOLZTNW/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":"2026-08-13T07:12:40Z","links":{"resolver":"https://pith.science/pith/3PGRP2FFOB44BDHLLPXEOLZTNW","bundle":"https://pith.science/pith/3PGRP2FFOB44BDHLLPXEOLZTNW/bundle.json","state":"https://pith.science/pith/3PGRP2FFOB44BDHLLPXEOLZTNW/state.json","well_known_bundle":"https://pith.science/.well-known/pith/3PGRP2FFOB44BDHLLPXEOLZTNW/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:3PGRP2FFOB44BDHLLPXEOLZTNW","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":"794c44ceb8e6590dcec6d2363c20ca67d1a9e373fb44c61f6dab1872dbb423ae","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2024-11-19T18:11:36Z","title_canon_sha256":"6acb00bfb4a1d42a7c4b20b54b5bb87016830fab6d252c5403d387423ed1b647"},"schema_version":"1.0","source":{"id":"2411.12701","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2411.12701","created_at":"2026-07-05T10:14:48Z"},{"alias_kind":"arxiv_version","alias_value":"2411.12701v3","created_at":"2026-07-05T10:14:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.12701","created_at":"2026-07-05T10:14:48Z"},{"alias_kind":"pith_short_12","alias_value":"3PGRP2FFOB44","created_at":"2026-07-05T10:14:48Z"},{"alias_kind":"pith_short_16","alias_value":"3PGRP2FFOB44BDHL","created_at":"2026-07-05T10:14:48Z"},{"alias_kind":"pith_short_8","alias_value":"3PGRP2FF","created_at":"2026-07-05T10:14:48Z"}],"graph_snapshots":[{"event_id":"sha256:c96479f7a2527048b8066075fa7f3ead71f414b7c3d947f8152240b3282ec781","target":"graph","created_at":"2026-07-05T10:14:48Z","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/2411.12701/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large Language Models (LLMs) are known to be vulnerable to backdoor attacks, where triggers embedded in poisoned samples can maliciously alter LLMs' behaviors. In this paper, we move beyond attacking LLMs and instead examine backdoor attacks through the novel lens of natural language explanations. Specifically, we leverage LLMs' generative capabilities to produce human-readable explanations for their decisions, enabling direct comparisons between explanations for clean and poisoned samples. Our results show that backdoored models produce coherent explanations for clean inputs but diverse and l","authors_text":"Huaizhi Ge, Qifan Wang, Ruixiang Tang, Yiming Li, Yongfeng Zhang","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2024-11-19T18:11:36Z","title":"When Backdoors Speak: Understanding LLM Backdoor Attacks Through Model-Generated Explanations"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.12701","kind":"arxiv","version":3},"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":"sha256:447cf7130aa8d5a234ca754875e0074efed1256dbf28f26cdec981c8482f0f1c","target":"record","created_at":"2026-07-05T10:14:48Z","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":"794c44ceb8e6590dcec6d2363c20ca67d1a9e373fb44c61f6dab1872dbb423ae","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2024-11-19T18:11:36Z","title_canon_sha256":"6acb00bfb4a1d42a7c4b20b54b5bb87016830fab6d252c5403d387423ed1b647"},"schema_version":"1.0","source":{"id":"2411.12701","kind":"arxiv","version":3}},"canonical_sha256":"dbcd17e8a57079c08ceb5bee472f336d84f335d530befa3979ec4958ee0ed1f3","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"dbcd17e8a57079c08ceb5bee472f336d84f335d530befa3979ec4958ee0ed1f3","first_computed_at":"2026-07-05T10:14:48.146895Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:14:48.146895Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"9L8CjndVbA/DBUtL9rh1Cvbe8KWdO4qjsD6ACXjD827+QI7YiNdynLPkIbnUp14Bm0T4+20KPGZwLgZK+ucdBA==","signature_status":"signed_v1","signed_at":"2026-07-05T10:14:48.147366Z","signed_message":"canonical_sha256_bytes"},"source_id":"2411.12701","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:447cf7130aa8d5a234ca754875e0074efed1256dbf28f26cdec981c8482f0f1c","sha256:c96479f7a2527048b8066075fa7f3ead71f414b7c3d947f8152240b3282ec781"],"state_sha256":"a68c78f6416e0834f0207ce2626119b07cf66080eea15e276f52e2efa3472a82"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"G1PbhrQJFnIe6Fivg09uW+qCnM8B7breb0EZ6LObIvKWMiz0CSLOI5Btm0LYMjv7JcJOmqfgYsclO9c+d7kgBg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-13T07:12:40.750605Z","bundle_sha256":"44cb0890f75404075b5fb4f9059e620d10475af5bbc4dd48e6e28bd8b2bbc708"}}