{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:XWAALZFBQ6SMTOPIACWLJ2FWAT","short_pith_number":"pith:XWAALZFB","canonical_record":{"source":{"id":"2407.01850","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-01T23:25:30Z","cross_cats_sorted":[],"title_canon_sha256":"9f940230e6542c160ecff4f67085ae79d99384e10c5ac55652e3495bd530a15b","abstract_canon_sha256":"0034ff972e28cf6315f92d1d316c8c610533ab6e05a550dcca3db711df749ad5"},"schema_version":"1.0"},"canonical_sha256":"bd8005e4a187a4c9b9e800acb4e8b604c59d7005c4c8e7fe53ab06854f30f3b3","source":{"kind":"arxiv","id":"2407.01850","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2407.01850","created_at":"2026-07-05T08:39:06Z"},{"alias_kind":"arxiv_version","alias_value":"2407.01850v1","created_at":"2026-07-05T08:39:06Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.01850","created_at":"2026-07-05T08:39:06Z"},{"alias_kind":"pith_short_12","alias_value":"XWAALZFBQ6SM","created_at":"2026-07-05T08:39:06Z"},{"alias_kind":"pith_short_16","alias_value":"XWAALZFBQ6SMTOPI","created_at":"2026-07-05T08:39:06Z"},{"alias_kind":"pith_short_8","alias_value":"XWAALZFB","created_at":"2026-07-05T08:39:06Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:XWAALZFBQ6SMTOPIACWLJ2FWAT","target":"record","payload":{"canonical_record":{"source":{"id":"2407.01850","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-01T23:25:30Z","cross_cats_sorted":[],"title_canon_sha256":"9f940230e6542c160ecff4f67085ae79d99384e10c5ac55652e3495bd530a15b","abstract_canon_sha256":"0034ff972e28cf6315f92d1d316c8c610533ab6e05a550dcca3db711df749ad5"},"schema_version":"1.0"},"canonical_sha256":"bd8005e4a187a4c9b9e800acb4e8b604c59d7005c4c8e7fe53ab06854f30f3b3","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:39:06.115842Z","signature_b64":"j0GVOMpMN6EvR8YZfoEQ7HuzT+5wBNEMiYEcFVuD6lW0QkYdO0b/DfrLbB4ac6GtHggkklha+4zS1km/qBuGAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bd8005e4a187a4c9b9e800acb4e8b604c59d7005c4c8e7fe53ab06854f30f3b3","last_reissued_at":"2026-07-05T08:39:06.115461Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:39:06.115461Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2407.01850","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":"2026-07-05T08:39:06Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"SBmTl0YglZ4toNr4pBq5DUcsWQpeNV9iqyVEkSes8JQyENm0ezTdEC1Y0j/GyykZwuWLfVAHW2k/wqBYB5ymBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-09T06:49:56.849168Z"},"content_sha256":"30735c5d5fe4899dae0c666d5d154a8fe1e8c56930eb99eef6585995e2586ef5","schema_version":"1.0","event_id":"sha256:30735c5d5fe4899dae0c666d5d154a8fe1e8c56930eb99eef6585995e2586ef5"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:XWAALZFBQ6SMTOPIACWLJ2FWAT","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Purple-teaming LLMs with Adversarial Defender Training","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Helen Meng, Jiawen Kang, Jingyan Zhou, Junan Li, Kun Li, Minda Hu, Xixin Wu","submitted_at":"2024-07-01T23:25:30Z","abstract_excerpt":"Existing efforts in safeguarding LLMs are limited in actively exposing the vulnerabilities of the target LLM and readily adapting to newly emerging safety risks. To address this, we present Purple-teaming LLMs with Adversarial Defender training (PAD), a pipeline designed to safeguard LLMs by novelly incorporating the red-teaming (attack) and blue-teaming (safety training) techniques. In PAD, we automatically collect conversational data that cover the vulnerabilities of an LLM around specific safety risks in a self-play manner, where the attacker aims to elicit unsafe responses and the defender"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.01850","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/2407.01850/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":"2026-07-05T08:39:06Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"GkUw4dhvD8CO9Xs0JTgcI9I++EKP4SlMwPvvj09kU9hIq6vlOuC74z53rLTeJQnrlnQjs2m1yT3jIhmGl9NUDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-09T06:49:56.849569Z"},"content_sha256":"78aa7c4e831c9d5df1255fe5dbce078e62f90db29e0072ddbdd70c8d19ff9aa6","schema_version":"1.0","event_id":"sha256:78aa7c4e831c9d5df1255fe5dbce078e62f90db29e0072ddbdd70c8d19ff9aa6"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/XWAALZFBQ6SMTOPIACWLJ2FWAT/bundle.json","state_url":"https://pith.science/pith/XWAALZFBQ6SMTOPIACWLJ2FWAT/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/XWAALZFBQ6SMTOPIACWLJ2FWAT/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":"2026-07-09T06:49:56Z","links":{"resolver":"https://pith.science/pith/XWAALZFBQ6SMTOPIACWLJ2FWAT","bundle":"https://pith.science/pith/XWAALZFBQ6SMTOPIACWLJ2FWAT/bundle.json","state":"https://pith.science/pith/XWAALZFBQ6SMTOPIACWLJ2FWAT/state.json","well_known_bundle":"https://pith.science/.well-known/pith/XWAALZFBQ6SMTOPIACWLJ2FWAT/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:XWAALZFBQ6SMTOPIACWLJ2FWAT","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":"0034ff972e28cf6315f92d1d316c8c610533ab6e05a550dcca3db711df749ad5","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-01T23:25:30Z","title_canon_sha256":"9f940230e6542c160ecff4f67085ae79d99384e10c5ac55652e3495bd530a15b"},"schema_version":"1.0","source":{"id":"2407.01850","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2407.01850","created_at":"2026-07-05T08:39:06Z"},{"alias_kind":"arxiv_version","alias_value":"2407.01850v1","created_at":"2026-07-05T08:39:06Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.01850","created_at":"2026-07-05T08:39:06Z"},{"alias_kind":"pith_short_12","alias_value":"XWAALZFBQ6SM","created_at":"2026-07-05T08:39:06Z"},{"alias_kind":"pith_short_16","alias_value":"XWAALZFBQ6SMTOPI","created_at":"2026-07-05T08:39:06Z"},{"alias_kind":"pith_short_8","alias_value":"XWAALZFB","created_at":"2026-07-05T08:39:06Z"}],"graph_snapshots":[{"event_id":"sha256:78aa7c4e831c9d5df1255fe5dbce078e62f90db29e0072ddbdd70c8d19ff9aa6","target":"graph","created_at":"2026-07-05T08:39:06Z","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/2407.01850/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Existing efforts in safeguarding LLMs are limited in actively exposing the vulnerabilities of the target LLM and readily adapting to newly emerging safety risks. To address this, we present Purple-teaming LLMs with Adversarial Defender training (PAD), a pipeline designed to safeguard LLMs by novelly incorporating the red-teaming (attack) and blue-teaming (safety training) techniques. In PAD, we automatically collect conversational data that cover the vulnerabilities of an LLM around specific safety risks in a self-play manner, where the attacker aims to elicit unsafe responses and the defender","authors_text":"Helen Meng, Jiawen Kang, Jingyan Zhou, Junan Li, Kun Li, Minda Hu, Xixin Wu","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-01T23:25:30Z","title":"Purple-teaming LLMs with Adversarial Defender Training"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.01850","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":"sha256:30735c5d5fe4899dae0c666d5d154a8fe1e8c56930eb99eef6585995e2586ef5","target":"record","created_at":"2026-07-05T08:39:06Z","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":"0034ff972e28cf6315f92d1d316c8c610533ab6e05a550dcca3db711df749ad5","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-01T23:25:30Z","title_canon_sha256":"9f940230e6542c160ecff4f67085ae79d99384e10c5ac55652e3495bd530a15b"},"schema_version":"1.0","source":{"id":"2407.01850","kind":"arxiv","version":1}},"canonical_sha256":"bd8005e4a187a4c9b9e800acb4e8b604c59d7005c4c8e7fe53ab06854f30f3b3","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"bd8005e4a187a4c9b9e800acb4e8b604c59d7005c4c8e7fe53ab06854f30f3b3","first_computed_at":"2026-07-05T08:39:06.115461Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:39:06.115461Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"j0GVOMpMN6EvR8YZfoEQ7HuzT+5wBNEMiYEcFVuD6lW0QkYdO0b/DfrLbB4ac6GtHggkklha+4zS1km/qBuGAQ==","signature_status":"signed_v1","signed_at":"2026-07-05T08:39:06.115842Z","signed_message":"canonical_sha256_bytes"},"source_id":"2407.01850","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:30735c5d5fe4899dae0c666d5d154a8fe1e8c56930eb99eef6585995e2586ef5","sha256:78aa7c4e831c9d5df1255fe5dbce078e62f90db29e0072ddbdd70c8d19ff9aa6"],"state_sha256":"7089d83d331905d3eb249f28d644ccde909af51b888c5b770f1ed27b38e04426"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"j1k49iZrC4PXJS9j1vUSG3ru+1uM3RuvhTDAiBI+/i5lKGUMyUcCPIOpE3LTpOMUrp3xWcjHU4urTw3u7MQlCA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-09T06:49:56.852204Z","bundle_sha256":"9ff05aa3d727518c13d026e98174009d1800aa76c3892abb94811d8bcc522b53"}}