{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:QA3EN4CRJOLR3DXCKRU2LPCS2I","short_pith_number":"pith:QA3EN4CR","schema_version":"1.0","canonical_sha256":"803646f0514b971d8ee25469a5bc52d21be40923907d94a6535bc7dc05978399","source":{"kind":"arxiv","id":"2402.07867","version":3},"attestation_state":"computed","paper":{"title":"PoisonedRAG: Knowledge Corruption Attacks to Retrieval-Augmented Generation of Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CR","authors_text":"Binghui Wang, Jinyuan Jia, Runpeng Geng, Wei Zou","submitted_at":"2024-02-12T18:28:36Z","abstract_excerpt":"Large language models (LLMs) have achieved remarkable success due to their exceptional generative capabilities. Despite their success, they also have inherent limitations such as a lack of up-to-date knowledge and hallucination. Retrieval-Augmented Generation (RAG) is a state-of-the-art technique to mitigate these limitations. The key idea of RAG is to ground the answer generation of an LLM on external knowledge retrieved from a knowledge database. Existing studies mainly focus on improving the accuracy or efficiency of RAG, leaving its security largely unexplored. We aim to bridge the gap in "},"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":"2402.07867","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-02-12T18:28:36Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"e1152312d667f3ff83e1dbd729b69a19525aae4ca626fd59a9f9f631c9a4939a","abstract_canon_sha256":"760914397ac407cb89f8620eff4a42915cb0093c24206d528a141a5c82cb7352"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:54:47.206109Z","signature_b64":"T4kNyxWAnhMqXVMR1XUIV4qmQN65FjT3wjkMoy6ZoHITVw5Z4jI6zJQneNe6iNYCRnA06j+/sporVyMG5hIuBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"803646f0514b971d8ee25469a5bc52d21be40923907d94a6535bc7dc05978399","last_reissued_at":"2026-07-05T08:54:47.205642Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:54:47.205642Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PoisonedRAG: Knowledge Corruption Attacks to Retrieval-Augmented Generation of Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CR","authors_text":"Binghui Wang, Jinyuan Jia, Runpeng Geng, Wei Zou","submitted_at":"2024-02-12T18:28:36Z","abstract_excerpt":"Large language models (LLMs) have achieved remarkable success due to their exceptional generative capabilities. Despite their success, they also have inherent limitations such as a lack of up-to-date knowledge and hallucination. Retrieval-Augmented Generation (RAG) is a state-of-the-art technique to mitigate these limitations. The key idea of RAG is to ground the answer generation of an LLM on external knowledge retrieved from a knowledge database. Existing studies mainly focus on improving the accuracy or efficiency of RAG, leaving its security largely unexplored. We aim to bridge the gap in "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.07867","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/2402.07867/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":"2402.07867","created_at":"2026-07-05T08:54:47.205697+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.07867v3","created_at":"2026-07-05T08:54:47.205697+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.07867","created_at":"2026-07-05T08:54:47.205697+00:00"},{"alias_kind":"pith_short_12","alias_value":"QA3EN4CRJOLR","created_at":"2026-07-05T08:54:47.205697+00:00"},{"alias_kind":"pith_short_16","alias_value":"QA3EN4CRJOLR3DXC","created_at":"2026-07-05T08:54:47.205697+00:00"},{"alias_kind":"pith_short_8","alias_value":"QA3EN4CR","created_at":"2026-07-05T08:54:47.205697+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":37,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.24322","citing_title":"Securing LLM-Agent Long-Term Memory Against Poisoning: Non-Malleable, Origin-Bound Authority with Machine-Checked Guarantees","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2606.22560","citing_title":"Evidence-Bound Gateway-Path Provenance for Third-Party LLM Inference","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2606.21377","citing_title":"ARENA: An Architecture for Measuring the Transferability of Autonomous Cyber Defense","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2606.18356","citing_title":"SafeClawBench: Separating Semantic, Audit-Evidence, and Sandbox Harm in Tool-Using LLM Agents","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2606.12290","citing_title":"Selection Integrity for LLM Graph Memory: An Accumulability Criterion for Information-Flow-Blind Retrieval","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2606.10322","citing_title":"Game-Theoretic Multi-Agent Control for Robust Contextual Reasoning in LLMs","ref_index":16,"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":136,"is_internal_anchor":false},{"citing_arxiv_id":"2606.04109","citing_title":"Discourse-Role Labels as Presentation-Time Variables for Context Use in Language Models","ref_index":16,"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":50,"is_internal_anchor":false},{"citing_arxiv_id":"2606.31639","citing_title":"A Lifecycle and Application-Stack Survey of Large Language Model Vulnerabilities: Attacks, Risks, Defenses, and Open Problems","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2606.32002","citing_title":"Self-Study Reconsidered: The Hidden Fragility of Learning from Self-Generated QA","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27736","citing_title":"ToE: A Hierarchical and Explainable Claim Verification Framework with Dynamic Multi-source Evidence Retrieval and Aggregation","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27157","citing_title":"Detecting Is Not Resolving: The Monitoring Control Gap in Retrieval Augmented LLMs","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00914","citing_title":"Adversarial Feeds Steer LLM Agent Decisions Against Their Defaults","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2606.02643","citing_title":"Inference Cost Attacks for Retrieval-Augmented Large Language Models","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2506.04390","citing_title":"Through the Stealth Lens: Attention-Aware Defenses Against Poisoning in RAG","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2409.10102","citing_title":"Trustworthiness in Retrieval-Augmented Generation Systems: A Survey","ref_index":79,"is_internal_anchor":false},{"citing_arxiv_id":"2412.14113","citing_title":"Adversarial Hubness in Multi-Modal Retrieval","ref_index":94,"is_internal_anchor":false},{"citing_arxiv_id":"2505.11548","citing_title":"One Shot Dominance: Knowledge Poisoning Attack on Retrieval-Augmented Generation Systems","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2510.13842","citing_title":"ADMIT: Few-shot Knowledge Poisoning Attacks on RAG-based Fact Checking","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.16282","citing_title":"Taxonomy and Consistency Analysis of Safety Benchmarks for AI Agents","ref_index":71,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18133","citing_title":"An Empirical Study of Privacy Leakage Chains via Prompt Injection in Black-Box Chatbot Environments","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14473","citing_title":"Does RAG Know When Retrieval Is Wrong? Diagnosing Context Compliance under Knowledge Conflict","ref_index":19,"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":34,"is_internal_anchor":false},{"citing_arxiv_id":"2510.18333","citing_title":"Position: LLM Watermarking Should Align Stakeholders' Incentives for Practical Adoption","ref_index":70,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QA3EN4CRJOLR3DXCKRU2LPCS2I","json":"https://pith.science/pith/QA3EN4CRJOLR3DXCKRU2LPCS2I.json","graph_json":"https://pith.science/api/pith-number/QA3EN4CRJOLR3DXCKRU2LPCS2I/graph.json","events_json":"https://pith.science/api/pith-number/QA3EN4CRJOLR3DXCKRU2LPCS2I/events.json","paper":"https://pith.science/paper/QA3EN4CR"},"agent_actions":{"view_html":"https://pith.science/pith/QA3EN4CRJOLR3DXCKRU2LPCS2I","download_json":"https://pith.science/pith/QA3EN4CRJOLR3DXCKRU2LPCS2I.json","view_paper":"https://pith.science/paper/QA3EN4CR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.07867&json=true","fetch_graph":"https://pith.science/api/pith-number/QA3EN4CRJOLR3DXCKRU2LPCS2I/graph.json","fetch_events":"https://pith.science/api/pith-number/QA3EN4CRJOLR3DXCKRU2LPCS2I/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QA3EN4CRJOLR3DXCKRU2LPCS2I/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QA3EN4CRJOLR3DXCKRU2LPCS2I/action/storage_attestation","attest_author":"https://pith.science/pith/QA3EN4CRJOLR3DXCKRU2LPCS2I/action/author_attestation","sign_citation":"https://pith.science/pith/QA3EN4CRJOLR3DXCKRU2LPCS2I/action/citation_signature","submit_replication":"https://pith.science/pith/QA3EN4CRJOLR3DXCKRU2LPCS2I/action/replication_record"}},"created_at":"2026-07-05T08:54:47.205697+00:00","updated_at":"2026-07-05T08:54:47.205697+00:00"}