{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:WOJK4LFMASWZ6G3PWJZYRI2DY5","short_pith_number":"pith:WOJK4LFM","schema_version":"1.0","canonical_sha256":"b392ae2cac04ad9f1b6fb27388a343c74eb7381622a02e5cea26be429d95931b","source":{"kind":"arxiv","id":"2509.23694","version":5},"attestation_state":"computed","paper":{"title":"SafeSearch: Automated Red-Teaming of LLM-Based Search Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.CR"],"primary_cat":"cs.AI","authors_text":"Han Qiu, Hao Wang, Jianshuo Dong, Ke Xu, Minlie Huang, Sheng Guo, Tianwei Zhang, Xun Chen, Zhuotao Liu","submitted_at":"2025-09-28T07:05:17Z","abstract_excerpt":"Search agents connect LLMs to the Internet, enabling them to access broader and more up-to-date information. However, this also introduces a new threat surface: unreliable search results can mislead agents into producing unsafe outputs. Real-world incidents and our two in-the-wild observations show that such failures can occur in practice. To study this threat systematically, we propose SafeSearch, an automated red-teaming framework that is scalable, cost-efficient, and lightweight, enabling sandboxed safety evaluation of search agents. Using this, we generate 300 test cases spanning five risk"},"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":"2509.23694","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-09-28T07:05:17Z","cross_cats_sorted":["cs.CL","cs.CR"],"title_canon_sha256":"e288ab1f125f53882cd11b0059b7fe8f8337b2db02e7a4ef31bf0e4d3a84adba","abstract_canon_sha256":"5936ad941f3e963e9fa221b61db69dbfcacf1f45c309ce4fb206ae6670f73e86"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-29T02:05:36.478891Z","signature_b64":"+eNQPkdzVkLn+4PjNkE/SQPg3ktY1uLRaFq/KT0NghB4idKpO6nk/7LcXqVyDjo54X41qykC40KA/I7mpoX5Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b392ae2cac04ad9f1b6fb27388a343c74eb7381622a02e5cea26be429d95931b","last_reissued_at":"2026-05-29T02:05:36.478478Z","signature_status":"signed_v1","first_computed_at":"2026-05-29T02:05:36.478478Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SafeSearch: Automated Red-Teaming of LLM-Based Search Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.CR"],"primary_cat":"cs.AI","authors_text":"Han Qiu, Hao Wang, Jianshuo Dong, Ke Xu, Minlie Huang, Sheng Guo, Tianwei Zhang, Xun Chen, Zhuotao Liu","submitted_at":"2025-09-28T07:05:17Z","abstract_excerpt":"Search agents connect LLMs to the Internet, enabling them to access broader and more up-to-date information. However, this also introduces a new threat surface: unreliable search results can mislead agents into producing unsafe outputs. Real-world incidents and our two in-the-wild observations show that such failures can occur in practice. To study this threat systematically, we propose SafeSearch, an automated red-teaming framework that is scalable, cost-efficient, and lightweight, enabling sandboxed safety evaluation of search agents. Using this, we generate 300 test cases spanning five risk"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.23694","kind":"arxiv","version":5},"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/2509.23694/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":"2509.23694","created_at":"2026-05-29T02:05:36.478532+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.23694v5","created_at":"2026-05-29T02:05:36.478532+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.23694","created_at":"2026-05-29T02:05:36.478532+00:00"},{"alias_kind":"pith_short_12","alias_value":"WOJK4LFMASWZ","created_at":"2026-05-29T02:05:36.478532+00:00"},{"alias_kind":"pith_short_16","alias_value":"WOJK4LFMASWZ6G3P","created_at":"2026-05-29T02:05:36.478532+00:00"},{"alias_kind":"pith_short_8","alias_value":"WOJK4LFM","created_at":"2026-05-29T02:05:36.478532+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.07110","citing_title":"Securing Computer-Use Agents: A Unified Architecture-Lifecycle Framework for Deployment-Grounded Reliability","ref_index":152,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WOJK4LFMASWZ6G3PWJZYRI2DY5","json":"https://pith.science/pith/WOJK4LFMASWZ6G3PWJZYRI2DY5.json","graph_json":"https://pith.science/api/pith-number/WOJK4LFMASWZ6G3PWJZYRI2DY5/graph.json","events_json":"https://pith.science/api/pith-number/WOJK4LFMASWZ6G3PWJZYRI2DY5/events.json","paper":"https://pith.science/paper/WOJK4LFM"},"agent_actions":{"view_html":"https://pith.science/pith/WOJK4LFMASWZ6G3PWJZYRI2DY5","download_json":"https://pith.science/pith/WOJK4LFMASWZ6G3PWJZYRI2DY5.json","view_paper":"https://pith.science/paper/WOJK4LFM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.23694&json=true","fetch_graph":"https://pith.science/api/pith-number/WOJK4LFMASWZ6G3PWJZYRI2DY5/graph.json","fetch_events":"https://pith.science/api/pith-number/WOJK4LFMASWZ6G3PWJZYRI2DY5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WOJK4LFMASWZ6G3PWJZYRI2DY5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WOJK4LFMASWZ6G3PWJZYRI2DY5/action/storage_attestation","attest_author":"https://pith.science/pith/WOJK4LFMASWZ6G3PWJZYRI2DY5/action/author_attestation","sign_citation":"https://pith.science/pith/WOJK4LFMASWZ6G3PWJZYRI2DY5/action/citation_signature","submit_replication":"https://pith.science/pith/WOJK4LFMASWZ6G3PWJZYRI2DY5/action/replication_record"}},"created_at":"2026-05-29T02:05:36.478532+00:00","updated_at":"2026-05-29T02:05:36.478532+00:00"}