{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:VRCEDPHLMSAKJRUX5Q5P2WT2IX","short_pith_number":"pith:VRCEDPHL","schema_version":"1.0","canonical_sha256":"ac4441bceb6480a4c697ec3afd5a7a45d52293751775c1e83c36984ff025df33","source":{"kind":"arxiv","id":"2606.02643","version":1},"attestation_state":"computed","paper":{"title":"Inference Cost Attacks for Retrieval-Augmented Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.DB"],"primary_cat":"cs.CR","authors_text":"Chengliang Liu, Liangbo Ning, Wenqi Fan, Yujuan Ding","submitted_at":"2026-05-31T15:11:59Z","abstract_excerpt":"Retrieval-Augmented Generation (RAG)-enhanced LLM systems, while powerful, introduce substantial inference costs due to the inclusion of an extra multi-stage pipeline that dynamically retrieves and synthesizes information from external knowledge sources. This high operational cost exposes a critical vulnerability to Inference Cost Attacks (ICAs). However, existing ICAs often rely on the impractical assumption of direct prompt manipulation. We argue that a more feasible and potent threat to RAG-enhanced LLM systems arises from poisoning external knowledge bases (e.g., web knowledge from the Int"},"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":"2606.02643","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2026-05-31T15:11:59Z","cross_cats_sorted":["cs.AI","cs.DB"],"title_canon_sha256":"310bd5c32b12c14131cba694f56f04e3cdbdf2b2c062ace50d81f8e9446a6e70","abstract_canon_sha256":"5159b565e94a739df565a8345090e3a19de4b075e4b5e4a232dce05f0f0596ad"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-03T00:05:05.479560Z","signature_b64":"cAlwJVl/vyGLKrF8IZU/4G73Dqstq/gx9nZ2RcaaO8m7AT705Ujb/6tQOHjif5+TOBR5Z9XegJI304Z8uQUnAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ac4441bceb6480a4c697ec3afd5a7a45d52293751775c1e83c36984ff025df33","last_reissued_at":"2026-06-03T00:05:05.479142Z","signature_status":"signed_v1","first_computed_at":"2026-06-03T00:05:05.479142Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Inference Cost Attacks for Retrieval-Augmented Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.DB"],"primary_cat":"cs.CR","authors_text":"Chengliang Liu, Liangbo Ning, Wenqi Fan, Yujuan Ding","submitted_at":"2026-05-31T15:11:59Z","abstract_excerpt":"Retrieval-Augmented Generation (RAG)-enhanced LLM systems, while powerful, introduce substantial inference costs due to the inclusion of an extra multi-stage pipeline that dynamically retrieves and synthesizes information from external knowledge sources. This high operational cost exposes a critical vulnerability to Inference Cost Attacks (ICAs). However, existing ICAs often rely on the impractical assumption of direct prompt manipulation. We argue that a more feasible and potent threat to RAG-enhanced LLM systems arises from poisoning external knowledge bases (e.g., web knowledge from the Int"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.02643","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/2606.02643/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":"2606.02643","created_at":"2026-06-03T00:05:05.479201+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.02643v1","created_at":"2026-06-03T00:05:05.479201+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.02643","created_at":"2026-06-03T00:05:05.479201+00:00"},{"alias_kind":"pith_short_12","alias_value":"VRCEDPHLMSAK","created_at":"2026-06-03T00:05:05.479201+00:00"},{"alias_kind":"pith_short_16","alias_value":"VRCEDPHLMSAKJRUX","created_at":"2026-06-03T00:05:05.479201+00:00"},{"alias_kind":"pith_short_8","alias_value":"VRCEDPHL","created_at":"2026-06-03T00:05:05.479201+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VRCEDPHLMSAKJRUX5Q5P2WT2IX","json":"https://pith.science/pith/VRCEDPHLMSAKJRUX5Q5P2WT2IX.json","graph_json":"https://pith.science/api/pith-number/VRCEDPHLMSAKJRUX5Q5P2WT2IX/graph.json","events_json":"https://pith.science/api/pith-number/VRCEDPHLMSAKJRUX5Q5P2WT2IX/events.json","paper":"https://pith.science/paper/VRCEDPHL"},"agent_actions":{"view_html":"https://pith.science/pith/VRCEDPHLMSAKJRUX5Q5P2WT2IX","download_json":"https://pith.science/pith/VRCEDPHLMSAKJRUX5Q5P2WT2IX.json","view_paper":"https://pith.science/paper/VRCEDPHL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.02643&json=true","fetch_graph":"https://pith.science/api/pith-number/VRCEDPHLMSAKJRUX5Q5P2WT2IX/graph.json","fetch_events":"https://pith.science/api/pith-number/VRCEDPHLMSAKJRUX5Q5P2WT2IX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VRCEDPHLMSAKJRUX5Q5P2WT2IX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VRCEDPHLMSAKJRUX5Q5P2WT2IX/action/storage_attestation","attest_author":"https://pith.science/pith/VRCEDPHLMSAKJRUX5Q5P2WT2IX/action/author_attestation","sign_citation":"https://pith.science/pith/VRCEDPHLMSAKJRUX5Q5P2WT2IX/action/citation_signature","submit_replication":"https://pith.science/pith/VRCEDPHLMSAKJRUX5Q5P2WT2IX/action/replication_record"}},"created_at":"2026-06-03T00:05:05.479201+00:00","updated_at":"2026-06-03T00:05:05.479201+00:00"}