{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:WWOG7BNK6NB6T4WLEW7GAYPSVZ","short_pith_number":"pith:WWOG7BNK","schema_version":"1.0","canonical_sha256":"b59c6f85aaf343e9f2cb25be6061f2ae62cd8d58c8b5889b6c8e419b3cad1f7c","source":{"kind":"arxiv","id":"2510.15476","version":3},"attestation_state":"computed","paper":{"title":"SoK: Systematizing LLM Prompt Security: Taxonomies, Datasets, and Unified Evaluation of Attacks and Defenses","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Ali Arastehfard, Hanbin Hong, Heqing Huang, Jingyu Zhang, Nima Naderloui, Shenao Yan, Shuang Wu, Shuya Feng, Yuan Hong","submitted_at":"2025-10-17T09:38:54Z","abstract_excerpt":"Large Language Models (LLMs) are increasingly used as interfaces to information, code, and real-world services, making prompt-level security failures a practical concern. Although jailbreak attacks, defenses, datasets, and automated judgers have advanced rapidly, evaluation remains fragmented across threat models, access assumptions, cost budgets, datasets, and success criteria. This makes reported attack success rates and defense gains hard to compare. This SoK systematizes LLM prompt security across concepts, data, tooling, and measurement. We propose linked taxonomies for jailbreak attacks,"},"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":"2510.15476","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-10-17T09:38:54Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"d0e51901e8c96a2c29d7a2cf494d08b49246c8051b6b9f3bd6be922ee7c2fb70","abstract_canon_sha256":"fad9a3ae32ae6ee9d21e2a60d342aecbd40097be9b4ddc609dfcda68079461c2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-07T02:17:12.100223Z","signature_b64":"IHIxnPJign69XX7H+N3JBPaHADnoqvVrJQkS/5ZdWuE3nPFvgmz2aX47TOSPmJavc7yqQ1x8zwHldZ+OYZuSBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b59c6f85aaf343e9f2cb25be6061f2ae62cd8d58c8b5889b6c8e419b3cad1f7c","last_reissued_at":"2026-07-07T02:17:12.099279Z","signature_status":"signed_v1","first_computed_at":"2026-07-07T02:17:12.099279Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SoK: Systematizing LLM Prompt Security: Taxonomies, Datasets, and Unified Evaluation of Attacks and Defenses","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CR","authors_text":"Ali Arastehfard, Hanbin Hong, Heqing Huang, Jingyu Zhang, Nima Naderloui, Shenao Yan, Shuang Wu, Shuya Feng, Yuan Hong","submitted_at":"2025-10-17T09:38:54Z","abstract_excerpt":"Large Language Models (LLMs) are increasingly used as interfaces to information, code, and real-world services, making prompt-level security failures a practical concern. Although jailbreak attacks, defenses, datasets, and automated judgers have advanced rapidly, evaluation remains fragmented across threat models, access assumptions, cost budgets, datasets, and success criteria. This makes reported attack success rates and defense gains hard to compare. This SoK systematizes LLM prompt security across concepts, data, tooling, and measurement. We propose linked taxonomies for jailbreak attacks,"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.15476","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/2510.15476/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":"2510.15476","created_at":"2026-07-07T02:17:12.099417+00:00"},{"alias_kind":"arxiv_version","alias_value":"2510.15476v3","created_at":"2026-07-07T02:17:12.099417+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.15476","created_at":"2026-07-07T02:17:12.099417+00:00"},{"alias_kind":"pith_short_12","alias_value":"WWOG7BNK6NB6","created_at":"2026-07-07T02:17:12.099417+00:00"},{"alias_kind":"pith_short_16","alias_value":"WWOG7BNK6NB6T4WL","created_at":"2026-07-07T02:17:12.099417+00:00"},{"alias_kind":"pith_short_8","alias_value":"WWOG7BNK","created_at":"2026-07-07T02:17:12.099417+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2606.25533","citing_title":"Security and Privacy in Retrieval-Augmented Generation: Architectures, Threats, Defenses, and Future Directions for Building Trustworthy Systems","ref_index":17,"is_internal_anchor":true},{"citing_arxiv_id":"2606.15549","citing_title":"One Goal, Many Commands: Characterizing Denylist Fragility in AI Agents","ref_index":51,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WWOG7BNK6NB6T4WLEW7GAYPSVZ","json":"https://pith.science/pith/WWOG7BNK6NB6T4WLEW7GAYPSVZ.json","graph_json":"https://pith.science/api/pith-number/WWOG7BNK6NB6T4WLEW7GAYPSVZ/graph.json","events_json":"https://pith.science/api/pith-number/WWOG7BNK6NB6T4WLEW7GAYPSVZ/events.json","paper":"https://pith.science/paper/WWOG7BNK"},"agent_actions":{"view_html":"https://pith.science/pith/WWOG7BNK6NB6T4WLEW7GAYPSVZ","download_json":"https://pith.science/pith/WWOG7BNK6NB6T4WLEW7GAYPSVZ.json","view_paper":"https://pith.science/paper/WWOG7BNK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2510.15476&json=true","fetch_graph":"https://pith.science/api/pith-number/WWOG7BNK6NB6T4WLEW7GAYPSVZ/graph.json","fetch_events":"https://pith.science/api/pith-number/WWOG7BNK6NB6T4WLEW7GAYPSVZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WWOG7BNK6NB6T4WLEW7GAYPSVZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WWOG7BNK6NB6T4WLEW7GAYPSVZ/action/storage_attestation","attest_author":"https://pith.science/pith/WWOG7BNK6NB6T4WLEW7GAYPSVZ/action/author_attestation","sign_citation":"https://pith.science/pith/WWOG7BNK6NB6T4WLEW7GAYPSVZ/action/citation_signature","submit_replication":"https://pith.science/pith/WWOG7BNK6NB6T4WLEW7GAYPSVZ/action/replication_record"}},"created_at":"2026-07-07T02:17:12.099417+00:00","updated_at":"2026-07-07T02:17:12.099417+00:00"}