{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:4PB5LOYAEKXOWAF24JTEJMH5HO","short_pith_number":"pith:4PB5LOYA","canonical_record":{"source":{"id":"2401.16656","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-01-30T01:19:25Z","cross_cats_sorted":[],"title_canon_sha256":"a397ae88039a0c847d323642fbed796b658278d3c27d056bf520916e957d8488","abstract_canon_sha256":"c07df44865cd7813f77b0af8bb1ae082191ca6609f948a9d77cdb9a5765ec3df"},"schema_version":"1.0"},"canonical_sha256":"e3c3d5bb0022aeeb00bae26644b0fd3b9b21994dbe89245b5566c43f2eacea0b","source":{"kind":"arxiv","id":"2401.16656","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2401.16656","created_at":"2026-07-05T07:38:58Z"},{"alias_kind":"arxiv_version","alias_value":"2401.16656v1","created_at":"2026-07-05T07:38:58Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.16656","created_at":"2026-07-05T07:38:58Z"},{"alias_kind":"pith_short_12","alias_value":"4PB5LOYAEKXO","created_at":"2026-07-05T07:38:58Z"},{"alias_kind":"pith_short_16","alias_value":"4PB5LOYAEKXOWAF2","created_at":"2026-07-05T07:38:58Z"},{"alias_kind":"pith_short_8","alias_value":"4PB5LOYA","created_at":"2026-07-05T07:38:58Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:4PB5LOYAEKXOWAF24JTEJMH5HO","target":"record","payload":{"canonical_record":{"source":{"id":"2401.16656","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-01-30T01:19:25Z","cross_cats_sorted":[],"title_canon_sha256":"a397ae88039a0c847d323642fbed796b658278d3c27d056bf520916e957d8488","abstract_canon_sha256":"c07df44865cd7813f77b0af8bb1ae082191ca6609f948a9d77cdb9a5765ec3df"},"schema_version":"1.0"},"canonical_sha256":"e3c3d5bb0022aeeb00bae26644b0fd3b9b21994dbe89245b5566c43f2eacea0b","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:38:58.336663Z","signature_b64":"2BLgnxV6SkxiHN7liHOL4IRtLGU3RcSRLRWHlGLTqTB2+rorcxxekRo9exN/tWZGTtVnJlr1Pha9a12NWvrEAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e3c3d5bb0022aeeb00bae26644b0fd3b9b21994dbe89245b5566c43f2eacea0b","last_reissued_at":"2026-07-05T07:38:58.336228Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:38:58.336228Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2401.16656","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":"2026-07-05T07:38:58Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Ps6KLqA/dvGWDrSlc6/PIFmt+coSiMOOiTIreOTPcg/7mSHJYjchFOKOEKhfBwhVxGvswzW6NHSc4KBTDJPZBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T22:22:12.188354Z"},"content_sha256":"ea473a8cfd05bf3a343dbcf16543199dd2c7738ddf7cc0f22e7ddd57d8a86ce9","schema_version":"1.0","event_id":"sha256:ea473a8cfd05bf3a343dbcf16543199dd2c7738ddf7cc0f22e7ddd57d8a86ce9"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:4PB5LOYAEKXOWAF24JTEJMH5HO","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Gradient-Based Language Model Red Teaming","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ahmad Beirami, Carson Denison, Nevan Wichers","submitted_at":"2024-01-30T01:19:25Z","abstract_excerpt":"Red teaming is a common strategy for identifying weaknesses in generative language models (LMs), where adversarial prompts are produced that trigger an LM to generate unsafe responses. Red teaming is instrumental for both model alignment and evaluation, but is labor-intensive and difficult to scale when done by humans. In this paper, we present Gradient-Based Red Teaming (GBRT), a red teaming method for automatically generating diverse prompts that are likely to cause an LM to output unsafe responses. GBRT is a form of prompt learning, trained by scoring an LM response with a safety classifier"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.16656","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/2401.16656/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":"2026-07-05T07:38:58Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"RdKonUvKm6a/dymK3eaCVT5pjw9wBSmok5PxUm9k/mjK47eGsMt0yV8pH3VWY9dQ0MyBkKnKfT+bkRiaPFWmCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T22:22:12.188895Z"},"content_sha256":"84058775d27740f7ca18010fb76ea86329b48d7371dafccaeda698cbdab87ea0","schema_version":"1.0","event_id":"sha256:84058775d27740f7ca18010fb76ea86329b48d7371dafccaeda698cbdab87ea0"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/4PB5LOYAEKXOWAF24JTEJMH5HO/bundle.json","state_url":"https://pith.science/pith/4PB5LOYAEKXOWAF24JTEJMH5HO/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/4PB5LOYAEKXOWAF24JTEJMH5HO/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":"2026-08-09T22:22:12Z","links":{"resolver":"https://pith.science/pith/4PB5LOYAEKXOWAF24JTEJMH5HO","bundle":"https://pith.science/pith/4PB5LOYAEKXOWAF24JTEJMH5HO/bundle.json","state":"https://pith.science/pith/4PB5LOYAEKXOWAF24JTEJMH5HO/state.json","well_known_bundle":"https://pith.science/.well-known/pith/4PB5LOYAEKXOWAF24JTEJMH5HO/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:4PB5LOYAEKXOWAF24JTEJMH5HO","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":"c07df44865cd7813f77b0af8bb1ae082191ca6609f948a9d77cdb9a5765ec3df","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-01-30T01:19:25Z","title_canon_sha256":"a397ae88039a0c847d323642fbed796b658278d3c27d056bf520916e957d8488"},"schema_version":"1.0","source":{"id":"2401.16656","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2401.16656","created_at":"2026-07-05T07:38:58Z"},{"alias_kind":"arxiv_version","alias_value":"2401.16656v1","created_at":"2026-07-05T07:38:58Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.16656","created_at":"2026-07-05T07:38:58Z"},{"alias_kind":"pith_short_12","alias_value":"4PB5LOYAEKXO","created_at":"2026-07-05T07:38:58Z"},{"alias_kind":"pith_short_16","alias_value":"4PB5LOYAEKXOWAF2","created_at":"2026-07-05T07:38:58Z"},{"alias_kind":"pith_short_8","alias_value":"4PB5LOYA","created_at":"2026-07-05T07:38:58Z"}],"graph_snapshots":[{"event_id":"sha256:84058775d27740f7ca18010fb76ea86329b48d7371dafccaeda698cbdab87ea0","target":"graph","created_at":"2026-07-05T07:38:58Z","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/2401.16656/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Red teaming is a common strategy for identifying weaknesses in generative language models (LMs), where adversarial prompts are produced that trigger an LM to generate unsafe responses. Red teaming is instrumental for both model alignment and evaluation, but is labor-intensive and difficult to scale when done by humans. In this paper, we present Gradient-Based Red Teaming (GBRT), a red teaming method for automatically generating diverse prompts that are likely to cause an LM to output unsafe responses. GBRT is a form of prompt learning, trained by scoring an LM response with a safety classifier","authors_text":"Ahmad Beirami, Carson Denison, Nevan Wichers","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-01-30T01:19:25Z","title":"Gradient-Based Language Model Red Teaming"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.16656","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:ea473a8cfd05bf3a343dbcf16543199dd2c7738ddf7cc0f22e7ddd57d8a86ce9","target":"record","created_at":"2026-07-05T07:38:58Z","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":"c07df44865cd7813f77b0af8bb1ae082191ca6609f948a9d77cdb9a5765ec3df","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-01-30T01:19:25Z","title_canon_sha256":"a397ae88039a0c847d323642fbed796b658278d3c27d056bf520916e957d8488"},"schema_version":"1.0","source":{"id":"2401.16656","kind":"arxiv","version":1}},"canonical_sha256":"e3c3d5bb0022aeeb00bae26644b0fd3b9b21994dbe89245b5566c43f2eacea0b","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"e3c3d5bb0022aeeb00bae26644b0fd3b9b21994dbe89245b5566c43f2eacea0b","first_computed_at":"2026-07-05T07:38:58.336228Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:38:58.336228Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"2BLgnxV6SkxiHN7liHOL4IRtLGU3RcSRLRWHlGLTqTB2+rorcxxekRo9exN/tWZGTtVnJlr1Pha9a12NWvrEAQ==","signature_status":"signed_v1","signed_at":"2026-07-05T07:38:58.336663Z","signed_message":"canonical_sha256_bytes"},"source_id":"2401.16656","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ea473a8cfd05bf3a343dbcf16543199dd2c7738ddf7cc0f22e7ddd57d8a86ce9","sha256:84058775d27740f7ca18010fb76ea86329b48d7371dafccaeda698cbdab87ea0"],"state_sha256":"30a21750be094d891805804fd53dbada57883fc55d9269a92adcaf9957a73ba5"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"rDg1meEd5RWuDCyHfVR9GUO8oo8n8/YAXxWrcXOBV9RchB+K1WuCdufLzEO+l6D0tfCgHs1XklOGQR39+nefCQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-09T22:22:12.192348Z","bundle_sha256":"944edec30c6d8bc4fdffb3dbb7cc0743899a29f32f47f716ac5f048814d9610c"}}