{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:FQWXF2653CH6JGFHYZPSPCSZUH","short_pith_number":"pith:FQWXF265","schema_version":"1.0","canonical_sha256":"2c2d72ebddd88fe498a7c65f278a59a1ebba2093ef4e8a6895a2ccdefd2edfc0","source":{"kind":"arxiv","id":"2606.09590","version":1},"attestation_state":"computed","paper":{"title":"Clinically Grounded Privacy Evaluation of Medical LMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.CL","authors_text":"Ayin Vala, Emily Alsentzer, Jordan Li Cahoon, Lena Stempfle, Marzyeh Ghassemi, Nathaniel Hendrix, Sana Tonekaboni, Sasha Ronaghi, Vivian Utti","submitted_at":"2026-06-08T15:02:19Z","abstract_excerpt":"Medical language models (LMs) can memorize and reproduce protected health information, but privacy evaluations often focus on recovery of training text rather than disclosure under realistic threat models. We introduce a clinically grounded framework that evaluates leakage along a graded axis of adversarial access, ranging from publicly inferable demographics to leaked note fragments. At each tier, we measure verbatim memorization of patient-specific text and semantic leakage of sensitive diagnoses. Applying the framework to an LM pretrained on 378k clinical notes, we find that routine encount"},"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":"2606.09590","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-08T15:02:19Z","cross_cats_sorted":["cs.CR"],"title_canon_sha256":"efde45ae5771d02177e4a04d3110df84d5a6a0a1b42cd72d22a77d3b9968dbac","abstract_canon_sha256":"401e6e70e8022dc1c17506f2494aa316f448552195a0970247a737103b434319"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T02:08:57.734868Z","signature_b64":"RqfIuvv7xesw+w1xvhOcTY2+vppc+kxiCjzLt8K1JVCkkiHfzErI3cG78EhrQ2+iF5YbjfVgXvVewLF0aRBpCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2c2d72ebddd88fe498a7c65f278a59a1ebba2093ef4e8a6895a2ccdefd2edfc0","last_reissued_at":"2026-06-09T02:08:57.733806Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T02:08:57.733806Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Clinically Grounded Privacy Evaluation of Medical LMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.CL","authors_text":"Ayin Vala, Emily Alsentzer, Jordan Li Cahoon, Lena Stempfle, Marzyeh Ghassemi, Nathaniel Hendrix, Sana Tonekaboni, Sasha Ronaghi, Vivian Utti","submitted_at":"2026-06-08T15:02:19Z","abstract_excerpt":"Medical language models (LMs) can memorize and reproduce protected health information, but privacy evaluations often focus on recovery of training text rather than disclosure under realistic threat models. We introduce a clinically grounded framework that evaluates leakage along a graded axis of adversarial access, ranging from publicly inferable demographics to leaked note fragments. At each tier, we measure verbatim memorization of patient-specific text and semantic leakage of sensitive diagnoses. Applying the framework to an LM pretrained on 378k clinical notes, we find that routine encount"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.09590","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/2606.09590/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":"2606.09590","created_at":"2026-06-09T02:08:57.733938+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.09590v1","created_at":"2026-06-09T02:08:57.733938+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.09590","created_at":"2026-06-09T02:08:57.733938+00:00"},{"alias_kind":"pith_short_12","alias_value":"FQWXF2653CH6","created_at":"2026-06-09T02:08:57.733938+00:00"},{"alias_kind":"pith_short_16","alias_value":"FQWXF2653CH6JGFH","created_at":"2026-06-09T02:08:57.733938+00:00"},{"alias_kind":"pith_short_8","alias_value":"FQWXF265","created_at":"2026-06-09T02:08:57.733938+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/FQWXF2653CH6JGFHYZPSPCSZUH","json":"https://pith.science/pith/FQWXF2653CH6JGFHYZPSPCSZUH.json","graph_json":"https://pith.science/api/pith-number/FQWXF2653CH6JGFHYZPSPCSZUH/graph.json","events_json":"https://pith.science/api/pith-number/FQWXF2653CH6JGFHYZPSPCSZUH/events.json","paper":"https://pith.science/paper/FQWXF265"},"agent_actions":{"view_html":"https://pith.science/pith/FQWXF2653CH6JGFHYZPSPCSZUH","download_json":"https://pith.science/pith/FQWXF2653CH6JGFHYZPSPCSZUH.json","view_paper":"https://pith.science/paper/FQWXF265","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.09590&json=true","fetch_graph":"https://pith.science/api/pith-number/FQWXF2653CH6JGFHYZPSPCSZUH/graph.json","fetch_events":"https://pith.science/api/pith-number/FQWXF2653CH6JGFHYZPSPCSZUH/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FQWXF2653CH6JGFHYZPSPCSZUH/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FQWXF2653CH6JGFHYZPSPCSZUH/action/storage_attestation","attest_author":"https://pith.science/pith/FQWXF2653CH6JGFHYZPSPCSZUH/action/author_attestation","sign_citation":"https://pith.science/pith/FQWXF2653CH6JGFHYZPSPCSZUH/action/citation_signature","submit_replication":"https://pith.science/pith/FQWXF2653CH6JGFHYZPSPCSZUH/action/replication_record"}},"created_at":"2026-06-09T02:08:57.733938+00:00","updated_at":"2026-06-09T02:08:57.733938+00:00"}