{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:R5ZDF3WCLXP7TLW6HAM3KSN6ZY","short_pith_number":"pith:R5ZDF3WC","schema_version":"1.0","canonical_sha256":"8f7232eec25ddff9aede3819b549bece3fdcdf3325338448a1dac6a85197faa8","source":{"kind":"arxiv","id":"2403.00108","version":2},"attestation_state":"computed","paper":{"title":"LoRATK: LoRA Once, Backdoor Everywhere in the Share-and-Play Ecosystem","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CR","authors_text":"Hongyi Liu, Jiayi Yuan, Li Li, Minghao Tian, Mohsen Hariri, Rui Chen, Ruixiang Tang, Shaochen Zhong, Soo-Hyun Choi, Vipin Chaudhary, Xia Hu, Xintong Sun, Yu-Neng Chuang, Zhimeng Jiang, Zirui Liu","submitted_at":"2024-02-29T20:25:16Z","abstract_excerpt":"Finetuning LLMs with LoRA has gained significant popularity due to its simplicity and effectiveness. Often, users may even find pluggable, community-shared LoRAs to enhance their base models for a specific downstream task of interest; enjoying a powerful, efficient, yet customized LLM experience with negligible investment. However, this convenient share-and-play ecosystem also introduces a new attack surface, where attackers can distribute malicious LoRAs to a community eager to try out shared assets. Despite the high-risk potential, no prior art has comprehensively explored LoRA's attack surf"},"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":"2403.00108","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2024-02-29T20:25:16Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"d5216ccc80fff55344632502c8c9194667d4008c9c5d2b57e14fe0ed103547c2","abstract_canon_sha256":"2893ec9e1334b42a19204bd1db6d3eaf75fc1f7983ccd062cb6d9f522c8db7f7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:56:32.281154Z","signature_b64":"lHeABgYS+PiH5IZNBx5qFLXIoNHt7293ti/xDm7STSjHV0OFJraEWUItOFRLHVFdChquPbEXIPxIUUUzqznhAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8f7232eec25ddff9aede3819b549bece3fdcdf3325338448a1dac6a85197faa8","last_reissued_at":"2026-07-05T10:56:32.280663Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:56:32.280663Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LoRATK: LoRA Once, Backdoor Everywhere in the Share-and-Play Ecosystem","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CR","authors_text":"Hongyi Liu, Jiayi Yuan, Li Li, Minghao Tian, Mohsen Hariri, Rui Chen, Ruixiang Tang, Shaochen Zhong, Soo-Hyun Choi, Vipin Chaudhary, Xia Hu, Xintong Sun, Yu-Neng Chuang, Zhimeng Jiang, Zirui Liu","submitted_at":"2024-02-29T20:25:16Z","abstract_excerpt":"Finetuning LLMs with LoRA has gained significant popularity due to its simplicity and effectiveness. Often, users may even find pluggable, community-shared LoRAs to enhance their base models for a specific downstream task of interest; enjoying a powerful, efficient, yet customized LLM experience with negligible investment. However, this convenient share-and-play ecosystem also introduces a new attack surface, where attackers can distribute malicious LoRAs to a community eager to try out shared assets. Despite the high-risk potential, no prior art has comprehensively explored LoRA's attack surf"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.00108","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/2403.00108/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":"2403.00108","created_at":"2026-07-05T10:56:32.280722+00:00"},{"alias_kind":"arxiv_version","alias_value":"2403.00108v2","created_at":"2026-07-05T10:56:32.280722+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.00108","created_at":"2026-07-05T10:56:32.280722+00:00"},{"alias_kind":"pith_short_12","alias_value":"R5ZDF3WCLXP7","created_at":"2026-07-05T10:56:32.280722+00:00"},{"alias_kind":"pith_short_16","alias_value":"R5ZDF3WCLXP7TLW6","created_at":"2026-07-05T10:56:32.280722+00:00"},{"alias_kind":"pith_short_8","alias_value":"R5ZDF3WC","created_at":"2026-07-05T10:56:32.280722+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.09151","citing_title":"Customization under Fire: Plugin Poisoning in Text-to-Image Ecosystem","ref_index":56,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07963","citing_title":"Shared Latent Structures Enable Unified Backdoor Detection and Mitigation in LLMs","ref_index":70,"is_internal_anchor":false},{"citing_arxiv_id":"2510.04265","citing_title":"Don't Pass@k: A Bayesian Framework for Large Language Model Evaluation","ref_index":62,"is_internal_anchor":false},{"citing_arxiv_id":"2408.07666","citing_title":"Model Merging in LLMs, MLLMs, and Beyond: Methods, Theories, Applications and Opportunities","ref_index":135,"is_internal_anchor":false},{"citing_arxiv_id":"2602.21977","citing_title":"When LoRA Betrays: Backdooring Text-to-Image Models by Masquerading as Benign Adapters","ref_index":24,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/R5ZDF3WCLXP7TLW6HAM3KSN6ZY","json":"https://pith.science/pith/R5ZDF3WCLXP7TLW6HAM3KSN6ZY.json","graph_json":"https://pith.science/api/pith-number/R5ZDF3WCLXP7TLW6HAM3KSN6ZY/graph.json","events_json":"https://pith.science/api/pith-number/R5ZDF3WCLXP7TLW6HAM3KSN6ZY/events.json","paper":"https://pith.science/paper/R5ZDF3WC"},"agent_actions":{"view_html":"https://pith.science/pith/R5ZDF3WCLXP7TLW6HAM3KSN6ZY","download_json":"https://pith.science/pith/R5ZDF3WCLXP7TLW6HAM3KSN6ZY.json","view_paper":"https://pith.science/paper/R5ZDF3WC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2403.00108&json=true","fetch_graph":"https://pith.science/api/pith-number/R5ZDF3WCLXP7TLW6HAM3KSN6ZY/graph.json","fetch_events":"https://pith.science/api/pith-number/R5ZDF3WCLXP7TLW6HAM3KSN6ZY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/R5ZDF3WCLXP7TLW6HAM3KSN6ZY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/R5ZDF3WCLXP7TLW6HAM3KSN6ZY/action/storage_attestation","attest_author":"https://pith.science/pith/R5ZDF3WCLXP7TLW6HAM3KSN6ZY/action/author_attestation","sign_citation":"https://pith.science/pith/R5ZDF3WCLXP7TLW6HAM3KSN6ZY/action/citation_signature","submit_replication":"https://pith.science/pith/R5ZDF3WCLXP7TLW6HAM3KSN6ZY/action/replication_record"}},"created_at":"2026-07-05T10:56:32.280722+00:00","updated_at":"2026-07-05T10:56:32.280722+00:00"}