{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:IX7OLQV74LGQ7HEAOBA2Y2PBBB","short_pith_number":"pith:IX7OLQV7","schema_version":"1.0","canonical_sha256":"45fee5c2bfe2cd0f9c807041ac69e1085cd33acf5c6c4ad08600cb0dc974d1f8","source":{"kind":"arxiv","id":"2406.00083","version":2},"attestation_state":"computed","paper":{"title":"BadRAG: Identifying Vulnerabilities in Retrieval Augmented Generation of Large Language Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.IR","cs.LG"],"primary_cat":"cs.CR","authors_text":"Fei Liu, Jiaqi Xue, Mengxin Zheng, Qian Lou, Xun Chen, Yebowen Hu","submitted_at":"2024-06-03T02:25:33Z","abstract_excerpt":"Large Language Models (LLMs) are constrained by outdated information and a tendency to generate incorrect data, commonly referred to as \"hallucinations.\" Retrieval-Augmented Generation (RAG) addresses these limitations by combining the strengths of retrieval-based methods and generative models. This approach involves retrieving relevant information from a large, up-to-date dataset and using it to enhance the generation process, leading to more accurate and contextually appropriate responses. Despite its benefits, RAG introduces a new attack surface for LLMs, particularly because RAG databases "},"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":"2406.00083","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CR","submitted_at":"2024-06-03T02:25:33Z","cross_cats_sorted":["cs.AI","cs.CL","cs.IR","cs.LG"],"title_canon_sha256":"7c80e9b35cb6e9f4fdbc20423c1d299530ee0aad0cb809a9d010d448e8787373","abstract_canon_sha256":"2c5126ddf119fe0b6343a868feccbc294c084dcf4e71c565eb27fbcffe1982a4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:28:10.810085Z","signature_b64":"CqTEVMvgJa5eXzem6xZDTCYuu0O2WfOR9RhIzDLlzqNhJZEaV7HRq2jSQPjOU4HBB5ozgRcV5SbiuMPIciNjCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"45fee5c2bfe2cd0f9c807041ac69e1085cd33acf5c6c4ad08600cb0dc974d1f8","last_reissued_at":"2026-07-05T08:28:10.809544Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:28:10.809544Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"BadRAG: Identifying Vulnerabilities in Retrieval Augmented Generation of Large Language Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.IR","cs.LG"],"primary_cat":"cs.CR","authors_text":"Fei Liu, Jiaqi Xue, Mengxin Zheng, Qian Lou, Xun Chen, Yebowen Hu","submitted_at":"2024-06-03T02:25:33Z","abstract_excerpt":"Large Language Models (LLMs) are constrained by outdated information and a tendency to generate incorrect data, commonly referred to as \"hallucinations.\" Retrieval-Augmented Generation (RAG) addresses these limitations by combining the strengths of retrieval-based methods and generative models. This approach involves retrieving relevant information from a large, up-to-date dataset and using it to enhance the generation process, leading to more accurate and contextually appropriate responses. Despite its benefits, RAG introduces a new attack surface for LLMs, particularly because RAG databases "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.00083","kind":"arxiv","version":2},"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/2406.00083/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":"2406.00083","created_at":"2026-07-05T08:28:10.809615+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.00083v2","created_at":"2026-07-05T08:28:10.809615+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.00083","created_at":"2026-07-05T08:28:10.809615+00:00"},{"alias_kind":"pith_short_12","alias_value":"IX7OLQV74LGQ","created_at":"2026-07-05T08:28:10.809615+00:00"},{"alias_kind":"pith_short_16","alias_value":"IX7OLQV74LGQ7HEA","created_at":"2026-07-05T08:28:10.809615+00:00"},{"alias_kind":"pith_short_8","alias_value":"IX7OLQV7","created_at":"2026-07-05T08:28:10.809615+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":24,"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":24,"is_internal_anchor":false},{"citing_arxiv_id":"2606.23892","citing_title":"REALM: A Unified Red-Teaming Benchmark for Physical-World VLMs","ref_index":35,"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":8,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17443","citing_title":"Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2607.00012","citing_title":"PRA-RAG: Provably Robust Aggregation in Retrieval-Augmented Generation against Retrieval Corruption","ref_index":128,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07783","citing_title":"Evaluating RAG Reliability under Clean, Misleading, and Mixed Retrieval","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03354","citing_title":"ImageAuditor: Membership Inference Attack against Image-based Retrieval-Augmented Generation","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01212","citing_title":"DiscourseFlip: An Oblique Discourse-Level Opinion Manipulation Attack against Black-box Retrieval-Augmented Generation","ref_index":40,"is_internal_anchor":false},{"citing_arxiv_id":"2605.28074","citing_title":"SilentRetrieval: Hijacking Retrieval-Augmented Generation via Semantically-Preserving Adversarial Data Poisoning","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2409.10102","citing_title":"Trustworthiness in Retrieval-Augmented Generation Systems: A Survey","ref_index":83,"is_internal_anchor":false},{"citing_arxiv_id":"2412.14113","citing_title":"Adversarial Hubness in Multi-Modal Retrieval","ref_index":83,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22041","citing_title":"RADAR: Defending RAG Dynamically against Retrieval Corruption","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09033","citing_title":"ShadowMerge: A Novel Poisoning Attack on Graph-Based Agent Memory via Relation-Channel Conflicts","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2506.03535","citing_title":"Across Programming Language Silos: A Study on Cross-Lingual Retrieval-augmented Code Generation","ref_index":59,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01758","citing_title":"Catching the Infection Before It Spreads: Foresight-Guided Defense in Multi-Agent Systems","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09033","citing_title":"ShadowMerge: A Novel Poisoning Attack on Graph-Based Agent Memory via Relation-Channel Conflicts","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11996","citing_title":"BadSKP: Backdoor Attacks on Knowledge Graph-Enhanced LLMs with Soft Prompts","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09033","citing_title":"ShadowMerge: A Novel Poisoning Attack on Graph-Based Agent Memory via Relation-Channel Conflicts","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09822","citing_title":"Oracle Poisoning: Corrupting Knowledge Graphs to Weaponise AI Agent Reasoning","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03482","citing_title":"MEMSAD: Gradient-Coupled Anomaly Detection for Memory Poisoning in Retrieval-Augmented Agents","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01782","citing_title":"Needle-in-RAG: Prompt-Conditioned Character-Level Traceback of Poisoned Spans in Retrieved Evidence","ref_index":54,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01758","citing_title":"Catching the Infection Before It Spreads: Foresight-Guided Defense in Multi-Agent Systems","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01758","citing_title":"Catching the Infection Before It Spreads: Foresight-Guided Defense in Multi-Agent Systems","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03482","citing_title":"MEMSAD: Gradient-Coupled Anomaly Detection for Memory Poisoning in Retrieval-Augmented Agents","ref_index":14,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IX7OLQV74LGQ7HEAOBA2Y2PBBB","json":"https://pith.science/pith/IX7OLQV74LGQ7HEAOBA2Y2PBBB.json","graph_json":"https://pith.science/api/pith-number/IX7OLQV74LGQ7HEAOBA2Y2PBBB/graph.json","events_json":"https://pith.science/api/pith-number/IX7OLQV74LGQ7HEAOBA2Y2PBBB/events.json","paper":"https://pith.science/paper/IX7OLQV7"},"agent_actions":{"view_html":"https://pith.science/pith/IX7OLQV74LGQ7HEAOBA2Y2PBBB","download_json":"https://pith.science/pith/IX7OLQV74LGQ7HEAOBA2Y2PBBB.json","view_paper":"https://pith.science/paper/IX7OLQV7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.00083&json=true","fetch_graph":"https://pith.science/api/pith-number/IX7OLQV74LGQ7HEAOBA2Y2PBBB/graph.json","fetch_events":"https://pith.science/api/pith-number/IX7OLQV74LGQ7HEAOBA2Y2PBBB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IX7OLQV74LGQ7HEAOBA2Y2PBBB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IX7OLQV74LGQ7HEAOBA2Y2PBBB/action/storage_attestation","attest_author":"https://pith.science/pith/IX7OLQV74LGQ7HEAOBA2Y2PBBB/action/author_attestation","sign_citation":"https://pith.science/pith/IX7OLQV74LGQ7HEAOBA2Y2PBBB/action/citation_signature","submit_replication":"https://pith.science/pith/IX7OLQV74LGQ7HEAOBA2Y2PBBB/action/replication_record"}},"created_at":"2026-07-05T08:28:10.809615+00:00","updated_at":"2026-07-05T08:28:10.809615+00:00"}