{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:UJI2RGZWIU5OOFUBL373726NGW","short_pith_number":"pith:UJI2RGZW","schema_version":"1.0","canonical_sha256":"a251a89b36453ae716815effbfebcd35b070105ea82bca394a3a22ca573b50e7","source":{"kind":"arxiv","id":"2409.19798","version":2},"attestation_state":"computed","paper":{"title":"Membership Inference Attacks Cannot Prove that a Model Was Trained On Your Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.LG","authors_text":"Debeshee Das, Florian Tram\\`er, Gautam Kamath, Jie Zhang","submitted_at":"2024-09-29T21:49:32Z","abstract_excerpt":"We consider the problem of a training data proof, where a data creator or owner wants to demonstrate to a third party that some machine learning model was trained on their data. Training data proofs play a key role in recent lawsuits against foundation models trained on web-scale data. Many prior works suggest to instantiate training data proofs using membership inference attacks. We argue that this approach is fundamentally unsound: to provide convincing evidence, the data creator needs to demonstrate that their attack has a low false positive rate, i.e., that the attack's output is unlikely "},"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":"2409.19798","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-09-29T21:49:32Z","cross_cats_sorted":["cs.CR"],"title_canon_sha256":"5b0e0a15d973980fa06a7ee6d2efaa12f6168c97f31a71eebf60d96028692139","abstract_canon_sha256":"490359ed240890cb59b08a370097e6423cf68ec3980a150468faa545b132356b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:26:07.949813Z","signature_b64":"TR3GA2lRZYJfCueNK92qLTlS3lRQJx0Ad0OrGwir8tAFFWx3nbANL2uPo4n2CAvo/310tElFnIVO0AN6uEtFAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a251a89b36453ae716815effbfebcd35b070105ea82bca394a3a22ca573b50e7","last_reissued_at":"2026-07-05T10:26:07.949094Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:26:07.949094Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Membership Inference Attacks Cannot Prove that a Model Was Trained On Your Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.LG","authors_text":"Debeshee Das, Florian Tram\\`er, Gautam Kamath, Jie Zhang","submitted_at":"2024-09-29T21:49:32Z","abstract_excerpt":"We consider the problem of a training data proof, where a data creator or owner wants to demonstrate to a third party that some machine learning model was trained on their data. Training data proofs play a key role in recent lawsuits against foundation models trained on web-scale data. Many prior works suggest to instantiate training data proofs using membership inference attacks. We argue that this approach is fundamentally unsound: to provide convincing evidence, the data creator needs to demonstrate that their attack has a low false positive rate, i.e., that the attack's output is unlikely "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.19798","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/2409.19798/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":"2409.19798","created_at":"2026-07-05T10:26:07.949164+00:00"},{"alias_kind":"arxiv_version","alias_value":"2409.19798v2","created_at":"2026-07-05T10:26:07.949164+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.19798","created_at":"2026-07-05T10:26:07.949164+00:00"},{"alias_kind":"pith_short_12","alias_value":"UJI2RGZWIU5O","created_at":"2026-07-05T10:26:07.949164+00:00"},{"alias_kind":"pith_short_16","alias_value":"UJI2RGZWIU5OOFUB","created_at":"2026-07-05T10:26:07.949164+00:00"},{"alias_kind":"pith_short_8","alias_value":"UJI2RGZW","created_at":"2026-07-05T10:26:07.949164+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.24408","citing_title":"Natural Identifiers for Privacy and Data Audits in Large Language Models","ref_index":22,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UJI2RGZWIU5OOFUBL373726NGW","json":"https://pith.science/pith/UJI2RGZWIU5OOFUBL373726NGW.json","graph_json":"https://pith.science/api/pith-number/UJI2RGZWIU5OOFUBL373726NGW/graph.json","events_json":"https://pith.science/api/pith-number/UJI2RGZWIU5OOFUBL373726NGW/events.json","paper":"https://pith.science/paper/UJI2RGZW"},"agent_actions":{"view_html":"https://pith.science/pith/UJI2RGZWIU5OOFUBL373726NGW","download_json":"https://pith.science/pith/UJI2RGZWIU5OOFUBL373726NGW.json","view_paper":"https://pith.science/paper/UJI2RGZW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2409.19798&json=true","fetch_graph":"https://pith.science/api/pith-number/UJI2RGZWIU5OOFUBL373726NGW/graph.json","fetch_events":"https://pith.science/api/pith-number/UJI2RGZWIU5OOFUBL373726NGW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UJI2RGZWIU5OOFUBL373726NGW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UJI2RGZWIU5OOFUBL373726NGW/action/storage_attestation","attest_author":"https://pith.science/pith/UJI2RGZWIU5OOFUBL373726NGW/action/author_attestation","sign_citation":"https://pith.science/pith/UJI2RGZWIU5OOFUBL373726NGW/action/citation_signature","submit_replication":"https://pith.science/pith/UJI2RGZWIU5OOFUBL373726NGW/action/replication_record"}},"created_at":"2026-07-05T10:26:07.949164+00:00","updated_at":"2026-07-05T10:26:07.949164+00:00"}