{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:QJDY4Y3ZXGOO52CUBNCGKNP6RV","short_pith_number":"pith:QJDY4Y3Z","schema_version":"1.0","canonical_sha256":"82478e6379b99ceee8540b446535fe8d65d0765e56e6bff9dbd0e44a46f4197a","source":{"kind":"arxiv","id":"2202.03286","version":1},"attestation_state":"computed","paper":{"title":"Red Teaming Language Models with Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"One language model generates test cases to automatically uncover tens of thousands of harmful behaviors in another language model.","cross_cats":["cs.AI","cs.CR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Amelia Glaese, Ethan Perez, Francis Song, Geoffrey Irving, John Aslanides, Nat McAleese, Roman Ring, Saffron Huang, Trevor Cai","submitted_at":"2022-02-07T15:22:17Z","abstract_excerpt":"Language Models (LMs) often cannot be deployed because of their potential to harm users in hard-to-predict ways. Prior work identifies harmful behaviors before deployment by using human annotators to hand-write test cases. However, human annotation is expensive, limiting the number and diversity of test cases. In this work, we automatically find cases where a target LM behaves in a harmful way, by generating test cases (\"red teaming\") using another LM. We evaluate the target LM's replies to generated test questions using a classifier trained to detect offensive content, uncovering tens of thou"},"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":true,"formal_links_present":false},"canonical_record":{"source":{"id":"2202.03286","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-02-07T15:22:17Z","cross_cats_sorted":["cs.AI","cs.CR","cs.LG"],"title_canon_sha256":"bb80b8b6cea33a49c34048dc37f0a2a6ceb18513c8d0fdde3c17a667d6323388","abstract_canon_sha256":"1ee4e6d94955492a585c6de94a1450e146e1c56b12e6b5d11fda444ee1f08518"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:54:37.615987Z","signature_b64":"9+MNayff/1mayqLmadyNBxRI8Gzme3+kX95n1xqRdeExFBiStHvYt484Df642h/02kVE8C/Iv2HxG1no9cU7Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"82478e6379b99ceee8540b446535fe8d65d0765e56e6bff9dbd0e44a46f4197a","last_reissued_at":"2026-07-05T03:54:37.615530Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:54:37.615530Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Red Teaming Language Models with Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"One language model generates test cases to automatically uncover tens of thousands of harmful behaviors in another language model.","cross_cats":["cs.AI","cs.CR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Amelia Glaese, Ethan Perez, Francis Song, Geoffrey Irving, John Aslanides, Nat McAleese, Roman Ring, Saffron Huang, Trevor Cai","submitted_at":"2022-02-07T15:22:17Z","abstract_excerpt":"Language Models (LMs) often cannot be deployed because of their potential to harm users in hard-to-predict ways. Prior work identifies harmful behaviors before deployment by using human annotators to hand-write test cases. However, human annotation is expensive, limiting the number and diversity of test cases. In this work, we automatically find cases where a target LM behaves in a harmful way, by generating test cases (\"red teaming\") using another LM. We evaluate the target LM's replies to generated test questions using a classifier trained to detect offensive content, uncovering tens of thou"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"we automatically find cases where a target LM behaves in a harmful way, by generating test cases (red teaming) using another LM... uncovering tens of thousands of offensive replies in a 280B parameter LM chatbot.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The classifier trained to detect offensive content accurately identifies the relevant harms, and the LM-generated test cases are sufficiently diverse, difficult, and representative of real user interactions.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"One language model can generate diverse test cases to automatically uncover tens of thousands of harmful behaviors, including offensive replies and privacy leaks, in a large target language model.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"One language model generates test cases to automatically uncover tens of thousands of harmful behaviors in another language model.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"1ca28634210af0459666ed92db628c2347a7c24b040f890a43e22d1a79d77650"},"source":{"id":"2202.03286","kind":"arxiv","version":1},"verdict":{"id":"11e1f2f3-d1bb-4af8-ac57-f6d9722da3e3","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-11T20:52:33.524394Z","strongest_claim":"we automatically find cases where a target LM behaves in a harmful way, by generating test cases (red teaming) using another LM... uncovering tens of thousands of offensive replies in a 280B parameter LM chatbot.","one_line_summary":"One language model can generate diverse test cases to automatically uncover tens of thousands of harmful behaviors, including offensive replies and privacy leaks, in a large target language model.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The classifier trained to detect offensive content accurately identifies the relevant harms, and the LM-generated test cases are sufficiently diverse, difficult, and representative of real user interactions.","pith_extraction_headline":"One language model generates test cases to automatically uncover tens of thousands of harmful behaviors in another language model."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2202.03286/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":15,"sample":[{"doi":"","year":2019,"title":"In ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 7345–7349","work_id":"ad751b1b-3615-45fc-bb1b-ff25cdd3c222","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1901,"title":"In Advances in Neural Information Processing Systems, volume 33, pages 1877–1901","work_id":"2dc7b88c-3d81-4a2d-ba10-e9527d40e62d","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2019,"title":"Unsolved Problems in ML Safety","work_id":"a4b59a6c-1b80-4562-afd6-0e6d0126d3bc","ref_index":3,"cited_arxiv_id":"2109.13916","is_internal_anchor":true},{"doi":"","year":2020,"title":"Way Off-Policy Batch Deep Reinforcement Learning of Implicit Human Preferences in Dialog","work_id":"42fcaa3e-0409-481b-9dd5-9a2c3be8a383","ref_index":4,"cited_arxiv_id":"1907.00456","is_internal_anchor":false},{"doi":"","year":2016,"title":"In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing , pages 2122–2132, Austin, Texas","work_id":"f5ab341f-5b15-47d7-8f8f-a43b384b5be8","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":15,"snapshot_sha256":"d657602cd5a9cebddef5dd057608b12d3f1113aa668e147ceabed942b54f103a","internal_anchors":2},"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":"2202.03286","created_at":"2026-07-05T03:54:37.615605+00:00"},{"alias_kind":"arxiv_version","alias_value":"2202.03286v1","created_at":"2026-07-05T03:54:37.615605+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.03286","created_at":"2026-07-05T03:54:37.615605+00:00"},{"alias_kind":"pith_short_12","alias_value":"QJDY4Y3ZXGOO","created_at":"2026-07-05T03:54:37.615605+00:00"},{"alias_kind":"pith_short_16","alias_value":"QJDY4Y3ZXGOO52CU","created_at":"2026-07-05T03:54:37.615605+00:00"},{"alias_kind":"pith_short_8","alias_value":"QJDY4Y3Z","created_at":"2026-07-05T03:54:37.615605+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":98,"internal_anchor_count":72,"sample":[{"citing_arxiv_id":"2606.25476","citing_title":"A Red Teaming Framework for Large Language Models: A Case Study on Faithfulness Evaluation","ref_index":50,"is_internal_anchor":true},{"citing_arxiv_id":"2606.24589","citing_title":"AdversaBench: Automated LLM Red-Teaming with Multi-Judge Confirmation and Cross-Model Transferability","ref_index":2,"is_internal_anchor":true},{"citing_arxiv_id":"2606.24937","citing_title":"The Hitchhiker's Guide to Agentic AI: From Foundations to Systems","ref_index":168,"is_internal_anchor":true},{"citing_arxiv_id":"2606.22329","citing_title":"BabelJudge: Measuring LLM-as-a-Judge Reliability Across Languages and Agent Trajectories","ref_index":12,"is_internal_anchor":true},{"citing_arxiv_id":"2606.19887","citing_title":"FinRED: An Expert-Guided Benchmark Generation and Evaluation Framework for Financial LLM Red-Teaming","ref_index":19,"is_internal_anchor":true},{"citing_arxiv_id":"2606.16000","citing_title":"GRACE-DS: a Guarded Reward-guided Agent Correction Environment in Data Science","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"2607.02079","citing_title":"HaloGuard 1.0: An Open Weights Constitutional Classifier for Multilingual AI Safety","ref_index":8,"is_internal_anchor":true},{"citing_arxiv_id":"2607.01859","citing_title":"Safety Targeted Embedding Exploit via Refinement","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2606.09701","citing_title":"Learning to Attack and Defend: Adaptive Red Teaming of Language Models via GRPO","ref_index":17,"is_internal_anchor":true},{"citing_arxiv_id":"2606.04037","citing_title":"Toward Pre-Deployment Assurance for Enterprise AI Agents: Ontology-Grounded Simulation and Trust Certification","ref_index":20,"is_internal_anchor":true},{"citing_arxiv_id":"2606.02041","citing_title":"SentGuard: Sentence-Level Streaming Guardrails for Large Language Models","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"2606.01695","citing_title":"CANARY: Zero-Label Detection of Fine-Tuning Contamination in Language Models","ref_index":11,"is_internal_anchor":true},{"citing_arxiv_id":"2606.01317","citing_title":"SABER: Benchmarking Operational Safety of LLM Coding Agents in Stateful Project Workspaces","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"2606.00150","citing_title":"Persona Attack: Incremental Memory Injection Jailbreak Attack against Large Language Models","ref_index":3,"is_internal_anchor":true},{"citing_arxiv_id":"2606.31748","citing_title":"Addressing Over-Refusal in LLMs with Competing Rewards","ref_index":64,"is_internal_anchor":true},{"citing_arxiv_id":"2606.30661","citing_title":"Understanding Censorship in Large Language Models: From Mechanisms to Governance","ref_index":26,"is_internal_anchor":true},{"citing_arxiv_id":"2605.17986","citing_title":"LivePI: More Realistic Benchmarking of Agents Against Indirect Prompt Injection","ref_index":35,"is_internal_anchor":true},{"citing_arxiv_id":"2605.23080","citing_title":"The Attribution Contract: Feature Attribution for Generative Language Models","ref_index":26,"is_internal_anchor":true},{"citing_arxiv_id":"2605.27766","citing_title":"Got a Secret? LLM Agents Can't Keep It: Evaluating Privacy in Multi-Agent Systems","ref_index":29,"is_internal_anchor":true},{"citing_arxiv_id":"2605.28664","citing_title":"Activation Steering for Synthetic Data Generation: The Role of Diversity in Downstream Safety Detection","ref_index":38,"is_internal_anchor":true},{"citing_arxiv_id":"2605.29659","citing_title":"Opir: Efficient Multi-Task Safety Classification for Toxicity, Jailbreaks, Hate Speech, and Harmful Content","ref_index":30,"is_internal_anchor":true},{"citing_arxiv_id":"2606.00320","citing_title":"Adversarially Robust Control of Conditional Value-at-Risk via Rockafellar-Uryasev Conformal Inference","ref_index":9,"is_internal_anchor":true},{"citing_arxiv_id":"2606.00801","citing_title":"Quality-Diversity Evolution for Discovering Diverse Vulnerabilities in LLM Safety","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2606.26566","citing_title":"Adversarial Diffusion Across Modalities: A Fusion Survey of Attacks, Defenses, and Evaluation for Text, Vision, and Vision-Language Models","ref_index":53,"is_internal_anchor":true},{"citing_arxiv_id":"2605.23080","citing_title":"The Attribution Contract: Feature Attribution for Generative Language Models","ref_index":26,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QJDY4Y3ZXGOO52CUBNCGKNP6RV","json":"https://pith.science/pith/QJDY4Y3ZXGOO52CUBNCGKNP6RV.json","graph_json":"https://pith.science/api/pith-number/QJDY4Y3ZXGOO52CUBNCGKNP6RV/graph.json","events_json":"https://pith.science/api/pith-number/QJDY4Y3ZXGOO52CUBNCGKNP6RV/events.json","paper":"https://pith.science/paper/QJDY4Y3Z"},"agent_actions":{"view_html":"https://pith.science/pith/QJDY4Y3ZXGOO52CUBNCGKNP6RV","download_json":"https://pith.science/pith/QJDY4Y3ZXGOO52CUBNCGKNP6RV.json","view_paper":"https://pith.science/paper/QJDY4Y3Z","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2202.03286&json=true","fetch_graph":"https://pith.science/api/pith-number/QJDY4Y3ZXGOO52CUBNCGKNP6RV/graph.json","fetch_events":"https://pith.science/api/pith-number/QJDY4Y3ZXGOO52CUBNCGKNP6RV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QJDY4Y3ZXGOO52CUBNCGKNP6RV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QJDY4Y3ZXGOO52CUBNCGKNP6RV/action/storage_attestation","attest_author":"https://pith.science/pith/QJDY4Y3ZXGOO52CUBNCGKNP6RV/action/author_attestation","sign_citation":"https://pith.science/pith/QJDY4Y3ZXGOO52CUBNCGKNP6RV/action/citation_signature","submit_replication":"https://pith.science/pith/QJDY4Y3ZXGOO52CUBNCGKNP6RV/action/replication_record"}},"created_at":"2026-07-05T03:54:37.615605+00:00","updated_at":"2026-07-05T03:54:37.615605+00:00"}