{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:4BJYMROODYYCYMUFVQYE2OWJMC","short_pith_number":"pith:4BJYMROO","schema_version":"1.0","canonical_sha256":"e0538645ce1e302c3285ac304d3ac960a2764f6414d3560273f3f3946da90869","source":{"kind":"arxiv","id":"2605.24173","version":1},"attestation_state":"computed","paper":{"title":"Extracting Training Data from Diffusion Language Models via Infilling","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CR","cs.LG"],"primary_cat":"cs.CL","authors_text":"N. Asokan, Yihan Wang","submitted_at":"2026-05-22T19:46:08Z","abstract_excerpt":"Memorization in large language models has been studied almost exclusively through prefix-conditioned extraction, a natural choice for autoregressive models. However, diffusion language models (DLMs) can denoise masked tokens at arbitrary positions. Thus, prefix-only probing reveals only one facet of memorization in DLMs and significantly underestimates the risk of training-data extraction. In order to realistically model extractability of training data in DLMs, we introduce \\emph{infilling extraction}, a data-extraction protocol parameterized by an arbitrary binary mask that subsumes prefix-on"},"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":"2605.24173","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-22T19:46:08Z","cross_cats_sorted":["cs.AI","cs.CR","cs.LG"],"title_canon_sha256":"00e04ca86a49d879c156cb80eaf24510ad38e23ee3dff724a032d83ef491fe03","abstract_canon_sha256":"8282bd2ab5541920b28caeeceb3b87ad26edba91dea0ca6ea63f31e5f55156d3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T01:02:50.569683Z","signature_b64":"l9pfQvQsjcs0SYMIBJS44Hlej/AFwrK/4zAdZ+eT3q2fiZoGVR1oXYH3znaY91efuBWlKjnSSO5w6cxcJN4uAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e0538645ce1e302c3285ac304d3ac960a2764f6414d3560273f3f3946da90869","last_reissued_at":"2026-05-26T01:02:50.568819Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T01:02:50.568819Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Extracting Training Data from Diffusion Language Models via Infilling","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CR","cs.LG"],"primary_cat":"cs.CL","authors_text":"N. Asokan, Yihan Wang","submitted_at":"2026-05-22T19:46:08Z","abstract_excerpt":"Memorization in large language models has been studied almost exclusively through prefix-conditioned extraction, a natural choice for autoregressive models. However, diffusion language models (DLMs) can denoise masked tokens at arbitrary positions. Thus, prefix-only probing reveals only one facet of memorization in DLMs and significantly underestimates the risk of training-data extraction. In order to realistically model extractability of training data in DLMs, we introduce \\emph{infilling extraction}, a data-extraction protocol parameterized by an arbitrary binary mask that subsumes prefix-on"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.24173","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/2605.24173/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":"2605.24173","created_at":"2026-05-26T01:02:50.568964+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.24173v1","created_at":"2026-05-26T01:02:50.568964+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.24173","created_at":"2026-05-26T01:02:50.568964+00:00"},{"alias_kind":"pith_short_12","alias_value":"4BJYMROODYYC","created_at":"2026-05-26T01:02:50.568964+00:00"},{"alias_kind":"pith_short_16","alias_value":"4BJYMROODYYCYMUF","created_at":"2026-05-26T01:02:50.568964+00:00"},{"alias_kind":"pith_short_8","alias_value":"4BJYMROO","created_at":"2026-05-26T01:02:50.568964+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/4BJYMROODYYCYMUFVQYE2OWJMC","json":"https://pith.science/pith/4BJYMROODYYCYMUFVQYE2OWJMC.json","graph_json":"https://pith.science/api/pith-number/4BJYMROODYYCYMUFVQYE2OWJMC/graph.json","events_json":"https://pith.science/api/pith-number/4BJYMROODYYCYMUFVQYE2OWJMC/events.json","paper":"https://pith.science/paper/4BJYMROO"},"agent_actions":{"view_html":"https://pith.science/pith/4BJYMROODYYCYMUFVQYE2OWJMC","download_json":"https://pith.science/pith/4BJYMROODYYCYMUFVQYE2OWJMC.json","view_paper":"https://pith.science/paper/4BJYMROO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.24173&json=true","fetch_graph":"https://pith.science/api/pith-number/4BJYMROODYYCYMUFVQYE2OWJMC/graph.json","fetch_events":"https://pith.science/api/pith-number/4BJYMROODYYCYMUFVQYE2OWJMC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4BJYMROODYYCYMUFVQYE2OWJMC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4BJYMROODYYCYMUFVQYE2OWJMC/action/storage_attestation","attest_author":"https://pith.science/pith/4BJYMROODYYCYMUFVQYE2OWJMC/action/author_attestation","sign_citation":"https://pith.science/pith/4BJYMROODYYCYMUFVQYE2OWJMC/action/citation_signature","submit_replication":"https://pith.science/pith/4BJYMROODYYCYMUFVQYE2OWJMC/action/replication_record"}},"created_at":"2026-05-26T01:02:50.568964+00:00","updated_at":"2026-05-26T01:02:50.568964+00:00"}