{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:RZMTYGMOZUSLIDABHH2I3H73W2","short_pith_number":"pith:RZMTYGMO","schema_version":"1.0","canonical_sha256":"8e593c198ecd24b40c0139f48d9ffbb693c89c79b3e3492317e562f4a59c656b","source":{"kind":"arxiv","id":"2308.09662","version":3},"attestation_state":"computed","paper":{"title":"Red-Teaming Large Language Models using Chain of Utterances for Safety-Alignment","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Rishabh Bhardwaj, Soujanya Poria","submitted_at":"2023-08-18T16:27:04Z","abstract_excerpt":"Larger language models (LLMs) have taken the world by storm with their massive multi-tasking capabilities simply by optimizing over a next-word prediction objective. With the emergence of their properties and encoded knowledge, the risk of LLMs producing harmful outputs increases, making them unfit for scalable deployment for the public. In this work, we propose a new safety evaluation benchmark RED-EVAL that carries out red-teaming. We show that even widely deployed models are susceptible to the Chain of Utterances-based (CoU) prompting, jailbreaking closed source LLM-based systems such as GP"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2308.09662","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2023-08-18T16:27:04Z","cross_cats_sorted":[],"title_canon_sha256":"7ebe9ba113a47315ed478c7980b68950fad4342f5697f3b9481afd68be4a7ea5","abstract_canon_sha256":"f33188da23c4c7ad8893e3a7c5c95df5a291303494fa96c6df9753d3c1da541a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:46:07.488050Z","signature_b64":"kb4b4pzjo3Ca4g9TUqAjbawLliGlaLQElw/KsmrEtz02BjsICkFQPwtXhwl035br9OSMz7v1LxumMoeYSjctBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8e593c198ecd24b40c0139f48d9ffbb693c89c79b3e3492317e562f4a59c656b","last_reissued_at":"2026-07-05T06:46:07.487628Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:46:07.487628Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Red-Teaming Large Language Models using Chain of Utterances for Safety-Alignment","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Rishabh Bhardwaj, Soujanya Poria","submitted_at":"2023-08-18T16:27:04Z","abstract_excerpt":"Larger language models (LLMs) have taken the world by storm with their massive multi-tasking capabilities simply by optimizing over a next-word prediction objective. With the emergence of their properties and encoded knowledge, the risk of LLMs producing harmful outputs increases, making them unfit for scalable deployment for the public. In this work, we propose a new safety evaluation benchmark RED-EVAL that carries out red-teaming. We show that even widely deployed models are susceptible to the Chain of Utterances-based (CoU) prompting, jailbreaking closed source LLM-based systems such as GP"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2308.09662","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/2308.09662/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2308.09662","created_at":"2026-07-05T06:46:07.487683+00:00"},{"alias_kind":"arxiv_version","alias_value":"2308.09662v3","created_at":"2026-07-05T06:46:07.487683+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2308.09662","created_at":"2026-07-05T06:46:07.487683+00:00"},{"alias_kind":"pith_short_12","alias_value":"RZMTYGMOZUSL","created_at":"2026-07-05T06:46:07.487683+00:00"},{"alias_kind":"pith_short_16","alias_value":"RZMTYGMOZUSLIDAB","created_at":"2026-07-05T06:46:07.487683+00:00"},{"alias_kind":"pith_short_8","alias_value":"RZMTYGMO","created_at":"2026-07-05T06:46:07.487683+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":17,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.24166","citing_title":"Distributed Quality-Diversity Search for Toxicity in Large Language Models","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07335","citing_title":"Defending Jailbreak Attacks on Large Language Models via Manifold Trajectory Kinetics","ref_index":83,"is_internal_anchor":false},{"citing_arxiv_id":"2606.02530","citing_title":"SafeSteer: Localized On-Policy Distillation for Efficient Safety Alignment","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00686","citing_title":"Dialectics of Alignment: Harnessing Unsafe Knowledge for Dynamic Safety Routing","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2408.12935","citing_title":"AI Safety Landscape for Large Language Models: Taxonomy, State-of-the-art, and Future Directions","ref_index":64,"is_internal_anchor":false},{"citing_arxiv_id":"2503.02574","citing_title":"LLM-Safety Evaluations Lack Robustness","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2505.14226","citing_title":"Phonetic Perturbations Reveal Tokenizer-Rooted Safety Gaps in LLMs","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2601.17887","citing_title":"When Personalization Legitimizes Risks: Uncovering Safety Vulnerabilities in Personalized Dialogue Agents","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19190","citing_title":"Going PLACES: Participatory Localized Red Teaming for Text-to-Image Safety in the Global South","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2509.10546","citing_title":"Learning to Conceal Risk: Controllable Multi-turn Red Teaming for LLMs in the Financial Domain","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2509.11206","citing_title":"Evalet: Evaluating Large Language Models through Functional Fragmentation","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2510.09689","citing_title":"When Search Goes Wrong: Red-Teaming Web-Augmented Large Language Models","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2601.20981","citing_title":"Diversifying Toxicity Search in Large Language Models Through Speciation","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2407.04295","citing_title":"Jailbreak Attacks and Defenses Against Large Language Models: A Survey","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04992","citing_title":"You Snooze, You Lose: Automatic Safety Alignment Restoration through Neural Weight Translation","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01687","citing_title":"MultiBreak: A Scalable and Diverse Multi-turn Jailbreak Benchmark for Evaluating LLM Safety","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04446","citing_title":"Misrouter: Exploiting Routing Mechanisms for Input-Only Attacks on Mixture-of-Experts LLMs","ref_index":4,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RZMTYGMOZUSLIDABHH2I3H73W2","json":"https://pith.science/pith/RZMTYGMOZUSLIDABHH2I3H73W2.json","graph_json":"https://pith.science/api/pith-number/RZMTYGMOZUSLIDABHH2I3H73W2/graph.json","events_json":"https://pith.science/api/pith-number/RZMTYGMOZUSLIDABHH2I3H73W2/events.json","paper":"https://pith.science/paper/RZMTYGMO"},"agent_actions":{"view_html":"https://pith.science/pith/RZMTYGMOZUSLIDABHH2I3H73W2","download_json":"https://pith.science/pith/RZMTYGMOZUSLIDABHH2I3H73W2.json","view_paper":"https://pith.science/paper/RZMTYGMO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2308.09662&json=true","fetch_graph":"https://pith.science/api/pith-number/RZMTYGMOZUSLIDABHH2I3H73W2/graph.json","fetch_events":"https://pith.science/api/pith-number/RZMTYGMOZUSLIDABHH2I3H73W2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RZMTYGMOZUSLIDABHH2I3H73W2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RZMTYGMOZUSLIDABHH2I3H73W2/action/storage_attestation","attest_author":"https://pith.science/pith/RZMTYGMOZUSLIDABHH2I3H73W2/action/author_attestation","sign_citation":"https://pith.science/pith/RZMTYGMOZUSLIDABHH2I3H73W2/action/citation_signature","submit_replication":"https://pith.science/pith/RZMTYGMOZUSLIDABHH2I3H73W2/action/replication_record"}},"created_at":"2026-07-05T06:46:07.487683+00:00","updated_at":"2026-07-05T06:46:07.487683+00:00"}