{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:N4FIMP7NOLK4NYDUP2PNVLOV5B","short_pith_number":"pith:N4FIMP7N","schema_version":"1.0","canonical_sha256":"6f0a863fed72d5c6e0747e9edaadd5e85bfaf3376758fc53ddfefc579192dd28","source":{"kind":"arxiv","id":"2401.00287","version":1},"attestation_state":"computed","paper":{"title":"The Art of Defending: A Systematic Evaluation and Analysis of LLM Defense Strategies on Safety and Over-Defensiveness","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Agastya Seth, Chitta Baral, Neeraj Varshney, Pavel Dolin","submitted_at":"2023-12-30T17:37:06Z","abstract_excerpt":"As Large Language Models (LLMs) play an increasingly pivotal role in natural language processing applications, their safety concerns become critical areas of NLP research. This paper presents Safety and Over-Defensiveness Evaluation (SODE) benchmark: a collection of diverse safe and unsafe prompts with carefully designed evaluation methods that facilitate systematic evaluation, comparison, and analysis over 'safety' and 'over-defensiveness.' With SODE, we study a variety of LLM defense strategies over multiple state-of-the-art LLMs, which reveals several interesting and important findings, suc"},"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":"2401.00287","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2023-12-30T17:37:06Z","cross_cats_sorted":[],"title_canon_sha256":"d85cd49660a99a3868f07161f662c65be295ea3da597b8df78f248151e24893e","abstract_canon_sha256":"01191c03fcca11a5f9df57c0e2b8639234165bd44e04d82f67c5e417e25491b9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:29:15.170047Z","signature_b64":"1h3/S8basoMhieeu/r34quySnWRD0XTdnbS1yMXg3P2zHo7XgNA9jHSd33Ge7bpbOcVHWu1c18sIQQtSynB4CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6f0a863fed72d5c6e0747e9edaadd5e85bfaf3376758fc53ddfefc579192dd28","last_reissued_at":"2026-07-05T07:29:15.169588Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:29:15.169588Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Art of Defending: A Systematic Evaluation and Analysis of LLM Defense Strategies on Safety and Over-Defensiveness","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Agastya Seth, Chitta Baral, Neeraj Varshney, Pavel Dolin","submitted_at":"2023-12-30T17:37:06Z","abstract_excerpt":"As Large Language Models (LLMs) play an increasingly pivotal role in natural language processing applications, their safety concerns become critical areas of NLP research. This paper presents Safety and Over-Defensiveness Evaluation (SODE) benchmark: a collection of diverse safe and unsafe prompts with carefully designed evaluation methods that facilitate systematic evaluation, comparison, and analysis over 'safety' and 'over-defensiveness.' With SODE, we study a variety of LLM defense strategies over multiple state-of-the-art LLMs, which reveals several interesting and important findings, suc"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.00287","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/2401.00287/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":"2401.00287","created_at":"2026-07-05T07:29:15.169654+00:00"},{"alias_kind":"arxiv_version","alias_value":"2401.00287v1","created_at":"2026-07-05T07:29:15.169654+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.00287","created_at":"2026-07-05T07:29:15.169654+00:00"},{"alias_kind":"pith_short_12","alias_value":"N4FIMP7NOLK4","created_at":"2026-07-05T07:29:15.169654+00:00"},{"alias_kind":"pith_short_16","alias_value":"N4FIMP7NOLK4NYDU","created_at":"2026-07-05T07:29:15.169654+00:00"},{"alias_kind":"pith_short_8","alias_value":"N4FIMP7N","created_at":"2026-07-05T07:29:15.169654+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.23817","citing_title":"System Prompt Extraction Attacks and Defenses in Large Language Models","ref_index":41,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/N4FIMP7NOLK4NYDUP2PNVLOV5B","json":"https://pith.science/pith/N4FIMP7NOLK4NYDUP2PNVLOV5B.json","graph_json":"https://pith.science/api/pith-number/N4FIMP7NOLK4NYDUP2PNVLOV5B/graph.json","events_json":"https://pith.science/api/pith-number/N4FIMP7NOLK4NYDUP2PNVLOV5B/events.json","paper":"https://pith.science/paper/N4FIMP7N"},"agent_actions":{"view_html":"https://pith.science/pith/N4FIMP7NOLK4NYDUP2PNVLOV5B","download_json":"https://pith.science/pith/N4FIMP7NOLK4NYDUP2PNVLOV5B.json","view_paper":"https://pith.science/paper/N4FIMP7N","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2401.00287&json=true","fetch_graph":"https://pith.science/api/pith-number/N4FIMP7NOLK4NYDUP2PNVLOV5B/graph.json","fetch_events":"https://pith.science/api/pith-number/N4FIMP7NOLK4NYDUP2PNVLOV5B/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/N4FIMP7NOLK4NYDUP2PNVLOV5B/action/timestamp_anchor","attest_storage":"https://pith.science/pith/N4FIMP7NOLK4NYDUP2PNVLOV5B/action/storage_attestation","attest_author":"https://pith.science/pith/N4FIMP7NOLK4NYDUP2PNVLOV5B/action/author_attestation","sign_citation":"https://pith.science/pith/N4FIMP7NOLK4NYDUP2PNVLOV5B/action/citation_signature","submit_replication":"https://pith.science/pith/N4FIMP7NOLK4NYDUP2PNVLOV5B/action/replication_record"}},"created_at":"2026-07-05T07:29:15.169654+00:00","updated_at":"2026-07-05T07:29:15.169654+00:00"}