{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:AEZDXEUXUVJBQZ35CKUDMDQVYZ","short_pith_number":"pith:AEZDXEUX","schema_version":"1.0","canonical_sha256":"01323b9297a55218677d12a8360e15c67c480e66d723bbb86bfd40959db49ec2","source":{"kind":"arxiv","id":"2007.02220","version":3},"attestation_state":"computed","paper":{"title":"You Autocomplete Me: Poisoning Vulnerabilities in Neural Code Completion","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.LG","cs.PL"],"primary_cat":"cs.CR","authors_text":"Congzheng Song, Eran Tromer, Roei Schuster, Vitaly Shmatikov","submitted_at":"2020-07-05T01:13:36Z","abstract_excerpt":"Code autocompletion is an integral feature of modern code editors and IDEs. The latest generation of autocompleters uses neural language models, trained on public open-source code repositories, to suggest likely (not just statically feasible) completions given the current context.\n  We demonstrate that neural code autocompleters are vulnerable to poisoning attacks. By adding a few specially-crafted files to the autocompleter's training corpus (data poisoning), or else by directly fine-tuning the autocompleter on these files (model poisoning), the attacker can influence its suggestions for atta"},"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":"2007.02220","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2020-07-05T01:13:36Z","cross_cats_sorted":["cs.CL","cs.LG","cs.PL"],"title_canon_sha256":"d471cef15811cacd354cdaa76561a934c13eb63831ed25af4105de97842f912b","abstract_canon_sha256":"3450eaa70a57ea1745168e52e4429cceadb8d593844c06aaf56b15f664d7a5e2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:41:32.460153Z","signature_b64":"XeH8WxvbV7f1+gDa4R2P/7PNj6y+oOwl+TyKn7/K8r5xSMo4YllGPspAgoBUK+TMR2jslKUs7YtOu0aK9K16Bg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"01323b9297a55218677d12a8360e15c67c480e66d723bbb86bfd40959db49ec2","last_reissued_at":"2026-07-05T01:41:32.459676Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:41:32.459676Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"You Autocomplete Me: Poisoning Vulnerabilities in Neural Code Completion","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.LG","cs.PL"],"primary_cat":"cs.CR","authors_text":"Congzheng Song, Eran Tromer, Roei Schuster, Vitaly Shmatikov","submitted_at":"2020-07-05T01:13:36Z","abstract_excerpt":"Code autocompletion is an integral feature of modern code editors and IDEs. The latest generation of autocompleters uses neural language models, trained on public open-source code repositories, to suggest likely (not just statically feasible) completions given the current context.\n  We demonstrate that neural code autocompleters are vulnerable to poisoning attacks. By adding a few specially-crafted files to the autocompleter's training corpus (data poisoning), or else by directly fine-tuning the autocompleter on these files (model poisoning), the attacker can influence its suggestions for atta"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2007.02220","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/2007.02220/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":"2007.02220","created_at":"2026-07-05T01:41:32.459733+00:00"},{"alias_kind":"arxiv_version","alias_value":"2007.02220v3","created_at":"2026-07-05T01:41:32.459733+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2007.02220","created_at":"2026-07-05T01:41:32.459733+00:00"},{"alias_kind":"pith_short_12","alias_value":"AEZDXEUXUVJB","created_at":"2026-07-05T01:41:32.459733+00:00"},{"alias_kind":"pith_short_16","alias_value":"AEZDXEUXUVJBQZ35","created_at":"2026-07-05T01:41:32.459733+00:00"},{"alias_kind":"pith_short_8","alias_value":"AEZDXEUX","created_at":"2026-07-05T01:41:32.459733+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.31246","citing_title":"BadBone: Backdoor Attacks Against Backbone Models in Visual Prompt Learning","ref_index":60,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AEZDXEUXUVJBQZ35CKUDMDQVYZ","json":"https://pith.science/pith/AEZDXEUXUVJBQZ35CKUDMDQVYZ.json","graph_json":"https://pith.science/api/pith-number/AEZDXEUXUVJBQZ35CKUDMDQVYZ/graph.json","events_json":"https://pith.science/api/pith-number/AEZDXEUXUVJBQZ35CKUDMDQVYZ/events.json","paper":"https://pith.science/paper/AEZDXEUX"},"agent_actions":{"view_html":"https://pith.science/pith/AEZDXEUXUVJBQZ35CKUDMDQVYZ","download_json":"https://pith.science/pith/AEZDXEUXUVJBQZ35CKUDMDQVYZ.json","view_paper":"https://pith.science/paper/AEZDXEUX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2007.02220&json=true","fetch_graph":"https://pith.science/api/pith-number/AEZDXEUXUVJBQZ35CKUDMDQVYZ/graph.json","fetch_events":"https://pith.science/api/pith-number/AEZDXEUXUVJBQZ35CKUDMDQVYZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AEZDXEUXUVJBQZ35CKUDMDQVYZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AEZDXEUXUVJBQZ35CKUDMDQVYZ/action/storage_attestation","attest_author":"https://pith.science/pith/AEZDXEUXUVJBQZ35CKUDMDQVYZ/action/author_attestation","sign_citation":"https://pith.science/pith/AEZDXEUXUVJBQZ35CKUDMDQVYZ/action/citation_signature","submit_replication":"https://pith.science/pith/AEZDXEUXUVJBQZ35CKUDMDQVYZ/action/replication_record"}},"created_at":"2026-07-05T01:41:32.459733+00:00","updated_at":"2026-07-05T01:41:32.459733+00:00"}