{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:KTOIJREHBKV3PNX7XQBVE7QNFA","short_pith_number":"pith:KTOIJREH","schema_version":"1.0","canonical_sha256":"54dc84c4870aabb7b6ffbc03527e0d2824ae7a5ae7dfa093fe1114ed4ba29dd5","source":{"kind":"arxiv","id":"2602.09319","version":3},"attestation_state":"computed","paper":{"title":"Benchmarking Knowledge-Extraction Attack and Defense on Retrieval-Augmented Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CR","authors_text":"Franck Dernoncourt, Haoyu Han, Li Ma, Mahantesh Halappanavar, Nesreen Ahmed, Ryan Rossi, Utkarsh Sahu, Yue Zhao, Yushun Dong, Yu Wang, Yu Zhang, Zhisheng Qi","submitted_at":"2026-02-10T01:27:46Z","abstract_excerpt":"Retrieval-Augmented Generation (RAG) has become a cornerstone of knowledge-intensive applications, including enterprise chatbots, healthcare assistants, and agentic memory management. However, recent studies show that knowledge-extraction attacks can recover sensitive knowledge-base content through maliciously crafted queries, raising serious intellectual property and privacy concerns. While prior work has explored individual attack and defense techniques, the research landscape remains fragmented, spanning heterogeneous retrieval embeddings, diverse generation models, and evaluations based on"},"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":"2602.09319","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2026-02-10T01:27:46Z","cross_cats_sorted":[],"title_canon_sha256":"c9e3a466651c78c317cf64e35cb1bb6517f750220cac6102870f7c3817709d88","abstract_canon_sha256":"cc559c579bd2bb14aa958151125994563a60e0640b64962b566a66a03a9d396a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-10T01:08:33.352981Z","signature_b64":"6vmeOn4NpKswrY6Ej+n7jbzwyH1VOYi/o05l0sfnoHjaLucb88HS+LKk3WINK82fG59AJNTaomftLm93cGYLCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"54dc84c4870aabb7b6ffbc03527e0d2824ae7a5ae7dfa093fe1114ed4ba29dd5","last_reissued_at":"2026-06-10T01:08:33.351795Z","signature_status":"signed_v1","first_computed_at":"2026-06-10T01:08:33.351795Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Benchmarking Knowledge-Extraction Attack and Defense on Retrieval-Augmented Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CR","authors_text":"Franck Dernoncourt, Haoyu Han, Li Ma, Mahantesh Halappanavar, Nesreen Ahmed, Ryan Rossi, Utkarsh Sahu, Yue Zhao, Yushun Dong, Yu Wang, Yu Zhang, Zhisheng Qi","submitted_at":"2026-02-10T01:27:46Z","abstract_excerpt":"Retrieval-Augmented Generation (RAG) has become a cornerstone of knowledge-intensive applications, including enterprise chatbots, healthcare assistants, and agentic memory management. However, recent studies show that knowledge-extraction attacks can recover sensitive knowledge-base content through maliciously crafted queries, raising serious intellectual property and privacy concerns. While prior work has explored individual attack and defense techniques, the research landscape remains fragmented, spanning heterogeneous retrieval embeddings, diverse generation models, and evaluations based on"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.09319","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/2602.09319/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":"2602.09319","created_at":"2026-06-10T01:08:33.351970+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.09319v3","created_at":"2026-06-10T01:08:33.351970+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.09319","created_at":"2026-06-10T01:08:33.351970+00:00"},{"alias_kind":"pith_short_12","alias_value":"KTOIJREHBKV3","created_at":"2026-06-10T01:08:33.351970+00:00"},{"alias_kind":"pith_short_16","alias_value":"KTOIJREHBKV3PNX7","created_at":"2026-06-10T01:08:33.351970+00:00"},{"alias_kind":"pith_short_8","alias_value":"KTOIJREH","created_at":"2026-06-10T01:08:33.351970+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.12827","citing_title":"GraphIP-Bench: How Hard Is It to Steal a Graph Neural Network, and Can We Stop It?","ref_index":19,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KTOIJREHBKV3PNX7XQBVE7QNFA","json":"https://pith.science/pith/KTOIJREHBKV3PNX7XQBVE7QNFA.json","graph_json":"https://pith.science/api/pith-number/KTOIJREHBKV3PNX7XQBVE7QNFA/graph.json","events_json":"https://pith.science/api/pith-number/KTOIJREHBKV3PNX7XQBVE7QNFA/events.json","paper":"https://pith.science/paper/KTOIJREH"},"agent_actions":{"view_html":"https://pith.science/pith/KTOIJREHBKV3PNX7XQBVE7QNFA","download_json":"https://pith.science/pith/KTOIJREHBKV3PNX7XQBVE7QNFA.json","view_paper":"https://pith.science/paper/KTOIJREH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.09319&json=true","fetch_graph":"https://pith.science/api/pith-number/KTOIJREHBKV3PNX7XQBVE7QNFA/graph.json","fetch_events":"https://pith.science/api/pith-number/KTOIJREHBKV3PNX7XQBVE7QNFA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KTOIJREHBKV3PNX7XQBVE7QNFA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KTOIJREHBKV3PNX7XQBVE7QNFA/action/storage_attestation","attest_author":"https://pith.science/pith/KTOIJREHBKV3PNX7XQBVE7QNFA/action/author_attestation","sign_citation":"https://pith.science/pith/KTOIJREHBKV3PNX7XQBVE7QNFA/action/citation_signature","submit_replication":"https://pith.science/pith/KTOIJREHBKV3PNX7XQBVE7QNFA/action/replication_record"}},"created_at":"2026-06-10T01:08:33.351970+00:00","updated_at":"2026-06-10T01:08:33.351970+00:00"}