{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:FQPCJ6D5YRVS6GVZ4XWWHVXKRV","short_pith_number":"pith:FQPCJ6D5","canonical_record":{"source":{"id":"2501.01830","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-01-03T14:30:14Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"21a07a1288aa2f23841b121e88a6a5f121a24f3d6c62bc0b60158cbc559f6a28","abstract_canon_sha256":"2896b620e386f56db4e1aa84d9bf687745ec760eb55ecc482632e369d9829702"},"schema_version":"1.0"},"canonical_sha256":"2c1e24f87dc46b2f1ab9e5ed63d6ea8d7083a6e3ec88da2f160a4f45e470462a","source":{"kind":"arxiv","id":"2501.01830","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.01830","created_at":"2026-07-05T09:56:41Z"},{"alias_kind":"arxiv_version","alias_value":"2501.01830v1","created_at":"2026-07-05T09:56:41Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.01830","created_at":"2026-07-05T09:56:41Z"},{"alias_kind":"pith_short_12","alias_value":"FQPCJ6D5YRVS","created_at":"2026-07-05T09:56:41Z"},{"alias_kind":"pith_short_16","alias_value":"FQPCJ6D5YRVS6GVZ","created_at":"2026-07-05T09:56:41Z"},{"alias_kind":"pith_short_8","alias_value":"FQPCJ6D5","created_at":"2026-07-05T09:56:41Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:FQPCJ6D5YRVS6GVZ4XWWHVXKRV","target":"record","payload":{"canonical_record":{"source":{"id":"2501.01830","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-01-03T14:30:14Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"21a07a1288aa2f23841b121e88a6a5f121a24f3d6c62bc0b60158cbc559f6a28","abstract_canon_sha256":"2896b620e386f56db4e1aa84d9bf687745ec760eb55ecc482632e369d9829702"},"schema_version":"1.0"},"canonical_sha256":"2c1e24f87dc46b2f1ab9e5ed63d6ea8d7083a6e3ec88da2f160a4f45e470462a","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:56:41.581273Z","signature_b64":"M+W/RTwavbDlN1kSvIVQt1J8GUW1LBfvKyf9aIv2rFWCOISxyY2ssufvj8xSlLXDW08L0bVswXS4T3GU995MBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2c1e24f87dc46b2f1ab9e5ed63d6ea8d7083a6e3ec88da2f160a4f45e470462a","last_reissued_at":"2026-07-05T09:56:41.580774Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:56:41.580774Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2501.01830","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":"2026-07-05T09:56:41Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"1d4h3usBhLYEgHk8iSEAbDOFgOBq3hqIqoN/XG0YMCxHwP94nrXNif6Xm3A7m3iAxcTYdWuyuxk5QUffop+aDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-14T17:06:17.287299Z"},"content_sha256":"321a3baf5f3b0fd4c63a69b23b1a0330813d24d4c7505ddbac6f8ede02c77d8c","schema_version":"1.0","event_id":"sha256:321a3baf5f3b0fd4c63a69b23b1a0330813d24d4c7505ddbac6f8ede02c77d8c"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:FQPCJ6D5YRVS6GVZ4XWWHVXKRV","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Auto-RT: Automatic Jailbreak Strategy Exploration for Red-Teaming Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CR","authors_text":"Ben He, Hongyu Lin, Huijia Zhu, Le Sun, Shuhen Zhou, Weiqiang Wang, Xianpei Han, Yanjiang Liu, Yaojie Lu","submitted_at":"2025-01-03T14:30:14Z","abstract_excerpt":"Automated red-teaming has become a crucial approach for uncovering vulnerabilities in large language models (LLMs). However, most existing methods focus on isolated safety flaws, limiting their ability to adapt to dynamic defenses and uncover complex vulnerabilities efficiently. To address this challenge, we propose Auto-RT, a reinforcement learning framework that automatically explores and optimizes complex attack strategies to effectively uncover security vulnerabilities through malicious queries. Specifically, we introduce two key mechanisms to reduce exploration complexity and improve stra"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.01830","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/2501.01830/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":"2026-07-05T09:56:41Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"QeIg/WkwpHyu7+WCrSd+4OZVqFNTBqpuXOicZhhCc+f5pgx2tb5g5PnoW5zs6nHJ3IuFwBGbxKBLIujLwDLCCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-14T17:06:17.287825Z"},"content_sha256":"86bd977b224cba4283da3a2567b8832138fabb884be7c27c86329218f067b182","schema_version":"1.0","event_id":"sha256:86bd977b224cba4283da3a2567b8832138fabb884be7c27c86329218f067b182"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/FQPCJ6D5YRVS6GVZ4XWWHVXKRV/bundle.json","state_url":"https://pith.science/pith/FQPCJ6D5YRVS6GVZ4XWWHVXKRV/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/FQPCJ6D5YRVS6GVZ4XWWHVXKRV/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":"2026-08-14T17:06:17Z","links":{"resolver":"https://pith.science/pith/FQPCJ6D5YRVS6GVZ4XWWHVXKRV","bundle":"https://pith.science/pith/FQPCJ6D5YRVS6GVZ4XWWHVXKRV/bundle.json","state":"https://pith.science/pith/FQPCJ6D5YRVS6GVZ4XWWHVXKRV/state.json","well_known_bundle":"https://pith.science/.well-known/pith/FQPCJ6D5YRVS6GVZ4XWWHVXKRV/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:FQPCJ6D5YRVS6GVZ4XWWHVXKRV","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":"2896b620e386f56db4e1aa84d9bf687745ec760eb55ecc482632e369d9829702","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-01-03T14:30:14Z","title_canon_sha256":"21a07a1288aa2f23841b121e88a6a5f121a24f3d6c62bc0b60158cbc559f6a28"},"schema_version":"1.0","source":{"id":"2501.01830","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.01830","created_at":"2026-07-05T09:56:41Z"},{"alias_kind":"arxiv_version","alias_value":"2501.01830v1","created_at":"2026-07-05T09:56:41Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.01830","created_at":"2026-07-05T09:56:41Z"},{"alias_kind":"pith_short_12","alias_value":"FQPCJ6D5YRVS","created_at":"2026-07-05T09:56:41Z"},{"alias_kind":"pith_short_16","alias_value":"FQPCJ6D5YRVS6GVZ","created_at":"2026-07-05T09:56:41Z"},{"alias_kind":"pith_short_8","alias_value":"FQPCJ6D5","created_at":"2026-07-05T09:56:41Z"}],"graph_snapshots":[{"event_id":"sha256:86bd977b224cba4283da3a2567b8832138fabb884be7c27c86329218f067b182","target":"graph","created_at":"2026-07-05T09:56:41Z","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/2501.01830/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Automated red-teaming has become a crucial approach for uncovering vulnerabilities in large language models (LLMs). However, most existing methods focus on isolated safety flaws, limiting their ability to adapt to dynamic defenses and uncover complex vulnerabilities efficiently. To address this challenge, we propose Auto-RT, a reinforcement learning framework that automatically explores and optimizes complex attack strategies to effectively uncover security vulnerabilities through malicious queries. Specifically, we introduce two key mechanisms to reduce exploration complexity and improve stra","authors_text":"Ben He, Hongyu Lin, Huijia Zhu, Le Sun, Shuhen Zhou, Weiqiang Wang, Xianpei Han, Yanjiang Liu, Yaojie Lu","cross_cats":["cs.AI","cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-01-03T14:30:14Z","title":"Auto-RT: Automatic Jailbreak Strategy Exploration for Red-Teaming Large Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.01830","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":"sha256:321a3baf5f3b0fd4c63a69b23b1a0330813d24d4c7505ddbac6f8ede02c77d8c","target":"record","created_at":"2026-07-05T09:56:41Z","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":"2896b620e386f56db4e1aa84d9bf687745ec760eb55ecc482632e369d9829702","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-01-03T14:30:14Z","title_canon_sha256":"21a07a1288aa2f23841b121e88a6a5f121a24f3d6c62bc0b60158cbc559f6a28"},"schema_version":"1.0","source":{"id":"2501.01830","kind":"arxiv","version":1}},"canonical_sha256":"2c1e24f87dc46b2f1ab9e5ed63d6ea8d7083a6e3ec88da2f160a4f45e470462a","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"2c1e24f87dc46b2f1ab9e5ed63d6ea8d7083a6e3ec88da2f160a4f45e470462a","first_computed_at":"2026-07-05T09:56:41.580774Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:56:41.580774Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"M+W/RTwavbDlN1kSvIVQt1J8GUW1LBfvKyf9aIv2rFWCOISxyY2ssufvj8xSlLXDW08L0bVswXS4T3GU995MBQ==","signature_status":"signed_v1","signed_at":"2026-07-05T09:56:41.581273Z","signed_message":"canonical_sha256_bytes"},"source_id":"2501.01830","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:321a3baf5f3b0fd4c63a69b23b1a0330813d24d4c7505ddbac6f8ede02c77d8c","sha256:86bd977b224cba4283da3a2567b8832138fabb884be7c27c86329218f067b182"],"state_sha256":"d348f200c11693e5fee7d4f6fce274794b42dea02bae7af1c171e66afb7f0282"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"lHGDc/HC9piwp6Eo8g80+uL1xIchddryDiiBvHeTCMx0Og+Sq0ZAVWvS4nMsylFLQHDq8lBX64GmzamWyoloDA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-14T17:06:17.297328Z","bundle_sha256":"5ab415d6c2deae05f27e91925b195256c3632716b8dcdfddeb7f4ccef7bc3d77"}}