{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:KWGZMEQENWDGTKZ2GJHCSZH2UJ","short_pith_number":"pith:KWGZMEQE","canonical_record":{"source":{"id":"2310.18603","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-28T06:11:07Z","cross_cats_sorted":[],"title_canon_sha256":"5cd8e8f9ee00d9b77563087e1b3f4e4adcc9c35fd94bbb0e6c379096f31a6ad4","abstract_canon_sha256":"3181ab87bb7fcf9630c22dace41f03ac8f5bc4dab3d904343edf4afb792735c0"},"schema_version":"1.0"},"canonical_sha256":"558d9612046d8669ab3a324e2964faa26aee4980385202561532ab8331c7fd28","source":{"kind":"arxiv","id":"2310.18603","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2310.18603","created_at":"2026-07-05T07:06:10Z"},{"alias_kind":"arxiv_version","alias_value":"2310.18603v1","created_at":"2026-07-05T07:06:10Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.18603","created_at":"2026-07-05T07:06:10Z"},{"alias_kind":"pith_short_12","alias_value":"KWGZMEQENWDG","created_at":"2026-07-05T07:06:10Z"},{"alias_kind":"pith_short_16","alias_value":"KWGZMEQENWDGTKZ2","created_at":"2026-07-05T07:06:10Z"},{"alias_kind":"pith_short_8","alias_value":"KWGZMEQE","created_at":"2026-07-05T07:06:10Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:KWGZMEQENWDGTKZ2GJHCSZH2UJ","target":"record","payload":{"canonical_record":{"source":{"id":"2310.18603","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-28T06:11:07Z","cross_cats_sorted":[],"title_canon_sha256":"5cd8e8f9ee00d9b77563087e1b3f4e4adcc9c35fd94bbb0e6c379096f31a6ad4","abstract_canon_sha256":"3181ab87bb7fcf9630c22dace41f03ac8f5bc4dab3d904343edf4afb792735c0"},"schema_version":"1.0"},"canonical_sha256":"558d9612046d8669ab3a324e2964faa26aee4980385202561532ab8331c7fd28","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:06:10.991685Z","signature_b64":"LsIw2sT52jHDzkPUEM3P+ihPJsEnFt559Mnr7UqsOBYjTlsideaDOH9oizoLNIn2xYf3FB5ifiIeQib55QTxDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"558d9612046d8669ab3a324e2964faa26aee4980385202561532ab8331c7fd28","last_reissued_at":"2026-07-05T07:06:10.991274Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:06:10.991274Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2310.18603","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":"2026-07-05T07:06:10Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"g4ffDlCpkRYPga45x4mWfxRDTTyUt1tOKS6jezom4SCZgAyN/b1UL/UyWgZ2HyYFpuojFHzuqH/PGi35F19KDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T04:55:28.422516Z"},"content_sha256":"ceaaccbb49bda608e8cbac17c43f9764f7228c56e819c6d96757806cdbac0255","schema_version":"1.0","event_id":"sha256:ceaaccbb49bda608e8cbac17c43f9764f7228c56e819c6d96757806cdbac0255"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:KWGZMEQENWDGTKZ2GJHCSZH2UJ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Large Language Models Are Better Adversaries: Exploring Generative Clean-Label Backdoor Attacks Against Text Classifiers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Daniel Lowd, Wencong You, Zayd Hammoudeh","submitted_at":"2023-10-28T06:11:07Z","abstract_excerpt":"Backdoor attacks manipulate model predictions by inserting innocuous triggers into training and test data. We focus on more realistic and more challenging clean-label attacks where the adversarial training examples are correctly labeled. Our attack, LLMBkd, leverages language models to automatically insert diverse style-based triggers into texts. We also propose a poison selection technique to improve the effectiveness of both LLMBkd as well as existing textual backdoor attacks. Lastly, we describe REACT, a baseline defense to mitigate backdoor attacks via antidote training examples. Our evalu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.18603","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/2310.18603/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":"2026-07-05T07:06:10Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"cTJXmGTwQF7bEvMLzEeQTgGCVxSnjoCTorqn3jlPT6NZ4aIVPaqD2w/quvRGOtK11/x+aEZM9e4mfhJi0bThAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-08T04:55:28.423186Z"},"content_sha256":"c8c81a53d7d89edc87185241760d49d094b4f354329b5bc613c3398fe98c3dea","schema_version":"1.0","event_id":"sha256:c8c81a53d7d89edc87185241760d49d094b4f354329b5bc613c3398fe98c3dea"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/KWGZMEQENWDGTKZ2GJHCSZH2UJ/bundle.json","state_url":"https://pith.science/pith/KWGZMEQENWDGTKZ2GJHCSZH2UJ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/KWGZMEQENWDGTKZ2GJHCSZH2UJ/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":"2026-08-08T04:55:28Z","links":{"resolver":"https://pith.science/pith/KWGZMEQENWDGTKZ2GJHCSZH2UJ","bundle":"https://pith.science/pith/KWGZMEQENWDGTKZ2GJHCSZH2UJ/bundle.json","state":"https://pith.science/pith/KWGZMEQENWDGTKZ2GJHCSZH2UJ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/KWGZMEQENWDGTKZ2GJHCSZH2UJ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:KWGZMEQENWDGTKZ2GJHCSZH2UJ","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":"3181ab87bb7fcf9630c22dace41f03ac8f5bc4dab3d904343edf4afb792735c0","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-28T06:11:07Z","title_canon_sha256":"5cd8e8f9ee00d9b77563087e1b3f4e4adcc9c35fd94bbb0e6c379096f31a6ad4"},"schema_version":"1.0","source":{"id":"2310.18603","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2310.18603","created_at":"2026-07-05T07:06:10Z"},{"alias_kind":"arxiv_version","alias_value":"2310.18603v1","created_at":"2026-07-05T07:06:10Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.18603","created_at":"2026-07-05T07:06:10Z"},{"alias_kind":"pith_short_12","alias_value":"KWGZMEQENWDG","created_at":"2026-07-05T07:06:10Z"},{"alias_kind":"pith_short_16","alias_value":"KWGZMEQENWDGTKZ2","created_at":"2026-07-05T07:06:10Z"},{"alias_kind":"pith_short_8","alias_value":"KWGZMEQE","created_at":"2026-07-05T07:06:10Z"}],"graph_snapshots":[{"event_id":"sha256:c8c81a53d7d89edc87185241760d49d094b4f354329b5bc613c3398fe98c3dea","target":"graph","created_at":"2026-07-05T07:06:10Z","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/2310.18603/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Backdoor attacks manipulate model predictions by inserting innocuous triggers into training and test data. We focus on more realistic and more challenging clean-label attacks where the adversarial training examples are correctly labeled. Our attack, LLMBkd, leverages language models to automatically insert diverse style-based triggers into texts. We also propose a poison selection technique to improve the effectiveness of both LLMBkd as well as existing textual backdoor attacks. Lastly, we describe REACT, a baseline defense to mitigate backdoor attacks via antidote training examples. Our evalu","authors_text":"Daniel Lowd, Wencong You, Zayd Hammoudeh","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-28T06:11:07Z","title":"Large Language Models Are Better Adversaries: Exploring Generative Clean-Label Backdoor Attacks Against Text Classifiers"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.18603","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:ceaaccbb49bda608e8cbac17c43f9764f7228c56e819c6d96757806cdbac0255","target":"record","created_at":"2026-07-05T07:06:10Z","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":"3181ab87bb7fcf9630c22dace41f03ac8f5bc4dab3d904343edf4afb792735c0","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-10-28T06:11:07Z","title_canon_sha256":"5cd8e8f9ee00d9b77563087e1b3f4e4adcc9c35fd94bbb0e6c379096f31a6ad4"},"schema_version":"1.0","source":{"id":"2310.18603","kind":"arxiv","version":1}},"canonical_sha256":"558d9612046d8669ab3a324e2964faa26aee4980385202561532ab8331c7fd28","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"558d9612046d8669ab3a324e2964faa26aee4980385202561532ab8331c7fd28","first_computed_at":"2026-07-05T07:06:10.991274Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:06:10.991274Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"LsIw2sT52jHDzkPUEM3P+ihPJsEnFt559Mnr7UqsOBYjTlsideaDOH9oizoLNIn2xYf3FB5ifiIeQib55QTxDA==","signature_status":"signed_v1","signed_at":"2026-07-05T07:06:10.991685Z","signed_message":"canonical_sha256_bytes"},"source_id":"2310.18603","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ceaaccbb49bda608e8cbac17c43f9764f7228c56e819c6d96757806cdbac0255","sha256:c8c81a53d7d89edc87185241760d49d094b4f354329b5bc613c3398fe98c3dea"],"state_sha256":"fb000f49684edcc26f1e941edcec97722dab4512d435080870da75798d321fe2"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"4mnijyGSpW1RWBZo8kzf/z7j1iQCIs+CZ7txZq0RvA/Hv7m2mGX8ZMKFzNcvEpc9Cn10YtQCMh2WSHkjBL4RAQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-08T04:55:28.430126Z","bundle_sha256":"39c292dd7e9d3664591d18d760bdb8b03da85af8d35ea537ad9d9eacb3242d0d"}}