{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:5UXZVVEIIZMPOJCJYFUALO6ZS7","short_pith_number":"pith:5UXZVVEI","schema_version":"1.0","canonical_sha256":"ed2f9ad4884658f72449c16805bbd997ff24cc9958c0d58b35721f8f0cc39306","source":{"kind":"arxiv","id":"2608.00144","version":1},"attestation_state":"computed","paper":{"title":"Leak It: A Probabilistic Approach to Training-Data Extraction from Black-Box Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.CR"],"primary_cat":"cs.LG","authors_text":"Victor Maricato","submitted_at":"2026-07-31T15:44:43Z","abstract_excerpt":"Membership inference (MIA) on language models is usually summarised by an aggregate ROC-AUC, but such evaluations are confounded: model-free blind baselines separate members from non-members from surface text alone. We study black-box, sampling-based training-data leakage through a probabilistic lens, treating N samples from p(.|x) as an estimate of the output distribution and casting leakage signals as functionals of it. We extend the blind-baseline critique into the sampling regime: on WikiMIA a blind bag-of-words classifier reaches AUC 0.97 (TPR 0.90 at 5% FPR) and sampling adds nothing, wh"},"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":"2608.00144","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-31T15:44:43Z","cross_cats_sorted":["cs.CL","cs.CR"],"title_canon_sha256":"cf9def1fb7be33597019208bbe14514be554560c1ed944f450deef9c5a77ec71","abstract_canon_sha256":"d72cc5b3fece72928462c260e5d5284ab71ce0d2017b0b84f49c7118a17a6318"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-04T00:33:03.138477Z","signature_b64":"y//Y0kAPxkxS21pJhqJl6M3VZO6Yb+lgSrkZUHbGKVB4HvF1tb7aj3r831j2t9zFogbBYinLhBIxXZ8RkKhTAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ed2f9ad4884658f72449c16805bbd997ff24cc9958c0d58b35721f8f0cc39306","last_reissued_at":"2026-08-04T00:33:03.137082Z","signature_status":"signed_v1","first_computed_at":"2026-08-04T00:33:03.137082Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Leak It: A Probabilistic Approach to Training-Data Extraction from Black-Box Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.CR"],"primary_cat":"cs.LG","authors_text":"Victor Maricato","submitted_at":"2026-07-31T15:44:43Z","abstract_excerpt":"Membership inference (MIA) on language models is usually summarised by an aggregate ROC-AUC, but such evaluations are confounded: model-free blind baselines separate members from non-members from surface text alone. We study black-box, sampling-based training-data leakage through a probabilistic lens, treating N samples from p(.|x) as an estimate of the output distribution and casting leakage signals as functionals of it. We extend the blind-baseline critique into the sampling regime: on WikiMIA a blind bag-of-words classifier reaches AUC 0.97 (TPR 0.90 at 5% FPR) and sampling adds nothing, wh"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.00144","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/2608.00144/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":"2608.00144","created_at":"2026-08-04T00:33:03.138124+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.00144v1","created_at":"2026-08-04T00:33:03.138124+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.00144","created_at":"2026-08-04T00:33:03.138124+00:00"},{"alias_kind":"pith_short_12","alias_value":"5UXZVVEIIZMP","created_at":"2026-08-04T00:33:03.138124+00:00"},{"alias_kind":"pith_short_16","alias_value":"5UXZVVEIIZMPOJCJ","created_at":"2026-08-04T00:33:03.138124+00:00"},{"alias_kind":"pith_short_8","alias_value":"5UXZVVEI","created_at":"2026-08-04T00:33:03.138124+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/5UXZVVEIIZMPOJCJYFUALO6ZS7","json":"https://pith.science/pith/5UXZVVEIIZMPOJCJYFUALO6ZS7.json","graph_json":"https://pith.science/api/pith-number/5UXZVVEIIZMPOJCJYFUALO6ZS7/graph.json","events_json":"https://pith.science/api/pith-number/5UXZVVEIIZMPOJCJYFUALO6ZS7/events.json","paper":"https://pith.science/paper/5UXZVVEI"},"agent_actions":{"view_html":"https://pith.science/pith/5UXZVVEIIZMPOJCJYFUALO6ZS7","download_json":"https://pith.science/pith/5UXZVVEIIZMPOJCJYFUALO6ZS7.json","view_paper":"https://pith.science/paper/5UXZVVEI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.00144&json=true","fetch_graph":"https://pith.science/api/pith-number/5UXZVVEIIZMPOJCJYFUALO6ZS7/graph.json","fetch_events":"https://pith.science/api/pith-number/5UXZVVEIIZMPOJCJYFUALO6ZS7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5UXZVVEIIZMPOJCJYFUALO6ZS7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5UXZVVEIIZMPOJCJYFUALO6ZS7/action/storage_attestation","attest_author":"https://pith.science/pith/5UXZVVEIIZMPOJCJYFUALO6ZS7/action/author_attestation","sign_citation":"https://pith.science/pith/5UXZVVEIIZMPOJCJYFUALO6ZS7/action/citation_signature","submit_replication":"https://pith.science/pith/5UXZVVEIIZMPOJCJYFUALO6ZS7/action/replication_record"}},"created_at":"2026-08-04T00:33:03.138124+00:00","updated_at":"2026-08-04T00:33:03.138124+00:00"}