{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:D65LYFG3EVJBSPTQDNIK3LVIYX","short_pith_number":"pith:D65LYFG3","schema_version":"1.0","canonical_sha256":"1fbabc14db2552193e701b50adaea8c5cd6bf3360268c4d45fd65790602d6fdd","source":{"kind":"arxiv","id":"2507.15042","version":1},"attestation_state":"computed","paper":{"title":"DeRAG: Black-box Adversarial Attacks on Multiple Retrieval-Augmented Generation Applications via Prompt Injection","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.IR"],"primary_cat":"cs.AI","authors_text":"Fang Yu, Jerry Wang","submitted_at":"2025-07-20T16:48:20Z","abstract_excerpt":"Adversarial prompt attacks can significantly alter the reliability of Retrieval-Augmented Generation (RAG) systems by re-ranking them to produce incorrect outputs. In this paper, we present a novel method that applies Differential Evolution (DE) to optimize adversarial prompt suffixes for RAG-based question answering. Our approach is gradient-free, treating the RAG pipeline as a black box and evolving a population of candidate suffixes to maximize the retrieval rank of a targeted incorrect document to be closer to real world scenarios. We conducted experiments on the BEIR QA datasets to evalua"},"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":"2507.15042","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2025-07-20T16:48:20Z","cross_cats_sorted":["cs.IR"],"title_canon_sha256":"a31a7b8567ee7e12ab10bac61fa0ff94b1f3f40183e7bfe2c2ad5c62a52a97c7","abstract_canon_sha256":"eba84616663836be272b131e43dff017a51e635096f8bc8999fc736c0c0bb8e8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:40:17.473379Z","signature_b64":"lv5eUVDxxYK0fbql0MBh0PBZ212c7jdxifKvmqQSZLcU42hs05Gww3T1/F0C65njG041FfCQphMygJA0OrX3DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1fbabc14db2552193e701b50adaea8c5cd6bf3360268c4d45fd65790602d6fdd","last_reissued_at":"2026-07-05T11:40:17.472920Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:40:17.472920Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DeRAG: Black-box Adversarial Attacks on Multiple Retrieval-Augmented Generation Applications via Prompt Injection","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.IR"],"primary_cat":"cs.AI","authors_text":"Fang Yu, Jerry Wang","submitted_at":"2025-07-20T16:48:20Z","abstract_excerpt":"Adversarial prompt attacks can significantly alter the reliability of Retrieval-Augmented Generation (RAG) systems by re-ranking them to produce incorrect outputs. In this paper, we present a novel method that applies Differential Evolution (DE) to optimize adversarial prompt suffixes for RAG-based question answering. Our approach is gradient-free, treating the RAG pipeline as a black box and evolving a population of candidate suffixes to maximize the retrieval rank of a targeted incorrect document to be closer to real world scenarios. We conducted experiments on the BEIR QA datasets to evalua"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.15042","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/2507.15042/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":"2507.15042","created_at":"2026-07-05T11:40:17.472972+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.15042v1","created_at":"2026-07-05T11:40:17.472972+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.15042","created_at":"2026-07-05T11:40:17.472972+00:00"},{"alias_kind":"pith_short_12","alias_value":"D65LYFG3EVJB","created_at":"2026-07-05T11:40:17.472972+00:00"},{"alias_kind":"pith_short_16","alias_value":"D65LYFG3EVJBSPTQ","created_at":"2026-07-05T11:40:17.472972+00:00"},{"alias_kind":"pith_short_8","alias_value":"D65LYFG3","created_at":"2026-07-05T11:40:17.472972+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"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":60,"is_internal_anchor":false},{"citing_arxiv_id":"2606.18310","citing_title":"Conflict-Aware Retriever Editing for Knowledge Injection Attacks on LLM-Based RAG Systems","ref_index":40,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/D65LYFG3EVJBSPTQDNIK3LVIYX","json":"https://pith.science/pith/D65LYFG3EVJBSPTQDNIK3LVIYX.json","graph_json":"https://pith.science/api/pith-number/D65LYFG3EVJBSPTQDNIK3LVIYX/graph.json","events_json":"https://pith.science/api/pith-number/D65LYFG3EVJBSPTQDNIK3LVIYX/events.json","paper":"https://pith.science/paper/D65LYFG3"},"agent_actions":{"view_html":"https://pith.science/pith/D65LYFG3EVJBSPTQDNIK3LVIYX","download_json":"https://pith.science/pith/D65LYFG3EVJBSPTQDNIK3LVIYX.json","view_paper":"https://pith.science/paper/D65LYFG3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.15042&json=true","fetch_graph":"https://pith.science/api/pith-number/D65LYFG3EVJBSPTQDNIK3LVIYX/graph.json","fetch_events":"https://pith.science/api/pith-number/D65LYFG3EVJBSPTQDNIK3LVIYX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/D65LYFG3EVJBSPTQDNIK3LVIYX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/D65LYFG3EVJBSPTQDNIK3LVIYX/action/storage_attestation","attest_author":"https://pith.science/pith/D65LYFG3EVJBSPTQDNIK3LVIYX/action/author_attestation","sign_citation":"https://pith.science/pith/D65LYFG3EVJBSPTQDNIK3LVIYX/action/citation_signature","submit_replication":"https://pith.science/pith/D65LYFG3EVJBSPTQDNIK3LVIYX/action/replication_record"}},"created_at":"2026-07-05T11:40:17.472972+00:00","updated_at":"2026-07-05T11:40:17.472972+00:00"}