{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:6OCIEMZ75ULDR2ERAXZ2WJ6PWO","short_pith_number":"pith:6OCIEMZ7","schema_version":"1.0","canonical_sha256":"f38482333fed1638e89105f3ab27cfb39caf241aa05270d225018868af785da4","source":{"kind":"arxiv","id":"2502.15010","version":2},"attestation_state":"computed","paper":{"title":"Obliviate: Efficient Unmemorization for Protecting Intellectual Property in Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Ahmed Salem, Mark Russinovich","submitted_at":"2025-02-20T20:02:56Z","abstract_excerpt":"Recent copyright agreements between AI companies and content creators underscore the need for fine-grained control over language models' ability to reproduce copyrighted text. Existing defenses-ranging from aggressive unlearning to simplistic output filters-either sacrifice model utility or inadequately address verbatim leakage. We introduce Obliviate, a lightweight post-training method that surgically suppresses exact reproduction of specified sequences while preserving semantic understanding. Obliviate first identifies memorized passages and then, for each target token, minimally adjusts the"},"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":"2502.15010","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-02-20T20:02:56Z","cross_cats_sorted":["cs.AI","cs.CR","cs.LG"],"title_canon_sha256":"dda6b2d73c4e635d820dc0700cdf394e841797fae24c298804e2a76803b3f230","abstract_canon_sha256":"c2ef6448618c5b210173f105645ecafeadb7edae85a81ee4d024fc80593c0bcf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:20:09.762176Z","signature_b64":"3vc0VCL+2ckxqx83GP6WpDwouqOyokG/A8PBhE+Vd2nF7RCB6YM6L5YJ/Rc7fUUJj65Vb4Y7JyRsZfi7mVbzAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f38482333fed1638e89105f3ab27cfb39caf241aa05270d225018868af785da4","last_reissued_at":"2026-07-05T11:20:09.761679Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:20:09.761679Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Obliviate: Efficient Unmemorization for Protecting Intellectual Property in Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Ahmed Salem, Mark Russinovich","submitted_at":"2025-02-20T20:02:56Z","abstract_excerpt":"Recent copyright agreements between AI companies and content creators underscore the need for fine-grained control over language models' ability to reproduce copyrighted text. Existing defenses-ranging from aggressive unlearning to simplistic output filters-either sacrifice model utility or inadequately address verbatim leakage. We introduce Obliviate, a lightweight post-training method that surgically suppresses exact reproduction of specified sequences while preserving semantic understanding. Obliviate first identifies memorized passages and then, for each target token, minimally adjusts the"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.15010","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/2502.15010/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":"2502.15010","created_at":"2026-07-05T11:20:09.761736+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.15010v2","created_at":"2026-07-05T11:20:09.761736+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.15010","created_at":"2026-07-05T11:20:09.761736+00:00"},{"alias_kind":"pith_short_12","alias_value":"6OCIEMZ75ULD","created_at":"2026-07-05T11:20:09.761736+00:00"},{"alias_kind":"pith_short_16","alias_value":"6OCIEMZ75ULDR2ER","created_at":"2026-07-05T11:20:09.761736+00:00"},{"alias_kind":"pith_short_8","alias_value":"6OCIEMZ7","created_at":"2026-07-05T11:20:09.761736+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.31250","citing_title":"Probing Stylistic Appropriation using Large Language Models: An Evaluation Framework for Copyright Infringement under EU Law","ref_index":61,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6OCIEMZ75ULDR2ERAXZ2WJ6PWO","json":"https://pith.science/pith/6OCIEMZ75ULDR2ERAXZ2WJ6PWO.json","graph_json":"https://pith.science/api/pith-number/6OCIEMZ75ULDR2ERAXZ2WJ6PWO/graph.json","events_json":"https://pith.science/api/pith-number/6OCIEMZ75ULDR2ERAXZ2WJ6PWO/events.json","paper":"https://pith.science/paper/6OCIEMZ7"},"agent_actions":{"view_html":"https://pith.science/pith/6OCIEMZ75ULDR2ERAXZ2WJ6PWO","download_json":"https://pith.science/pith/6OCIEMZ75ULDR2ERAXZ2WJ6PWO.json","view_paper":"https://pith.science/paper/6OCIEMZ7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.15010&json=true","fetch_graph":"https://pith.science/api/pith-number/6OCIEMZ75ULDR2ERAXZ2WJ6PWO/graph.json","fetch_events":"https://pith.science/api/pith-number/6OCIEMZ75ULDR2ERAXZ2WJ6PWO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6OCIEMZ75ULDR2ERAXZ2WJ6PWO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6OCIEMZ75ULDR2ERAXZ2WJ6PWO/action/storage_attestation","attest_author":"https://pith.science/pith/6OCIEMZ75ULDR2ERAXZ2WJ6PWO/action/author_attestation","sign_citation":"https://pith.science/pith/6OCIEMZ75ULDR2ERAXZ2WJ6PWO/action/citation_signature","submit_replication":"https://pith.science/pith/6OCIEMZ75ULDR2ERAXZ2WJ6PWO/action/replication_record"}},"created_at":"2026-07-05T11:20:09.761736+00:00","updated_at":"2026-07-05T11:20:09.761736+00:00"}