{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:NFIXUZEJOBK6HGSPYHMOEQY5Z6","short_pith_number":"pith:NFIXUZEJ","schema_version":"1.0","canonical_sha256":"69517a64897055e39a4fc1d8e2431dcfaf14f9a7d2a957a5eaebc05f0a1e0848","source":{"kind":"arxiv","id":"2410.02467","version":7},"attestation_state":"computed","paper":{"title":"SIDE: Surrogate Conditional Data Extraction from Diffusion Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CR","cs.CV"],"primary_cat":"cs.LG","authors_text":"Difan Zou, Shujie Wang, Xingjun Ma, Yunhao Chen","submitted_at":"2024-10-03T13:17:06Z","abstract_excerpt":"As diffusion probabilistic models (DPMs) become central to Generative AI (GenAI), understanding their memorization behavior is essential for evaluating risks such as data leakage, copyright infringement, and trustworthiness. While prior research finds conditional DPMs highly susceptible to data extraction attacks using explicit prompts, unconditional models are often assumed to be safe. We challenge this view by introducing \\textbf{Surrogate condItional Data Extraction (SIDE)}, a general framework that constructs data-driven surrogate conditions to enable targeted extraction from any DPM. Thro"},"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":"2410.02467","kind":"arxiv","version":7},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2024-10-03T13:17:06Z","cross_cats_sorted":["cs.CR","cs.CV"],"title_canon_sha256":"c9c2e068f6bbe64f0a4132dd25d85cbd2625ca9afc68f349255d4268facb558d","abstract_canon_sha256":"13eea3438272e220a6d57b3e077c4af4317fc563abe270b7fa4e336cbedbe77f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:46:36.371420Z","signature_b64":"8dh3Z79ag4eBmo2rNkP9Pb1ZcU2U6MDv7LKdZJFLiPGz/Eix17Sy1ZRHulghSgiisLhI4csqs3CNEuaOjLWpCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"69517a64897055e39a4fc1d8e2431dcfaf14f9a7d2a957a5eaebc05f0a1e0848","last_reissued_at":"2026-07-05T11:46:36.370880Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:46:36.370880Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SIDE: Surrogate Conditional Data Extraction from Diffusion Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CR","cs.CV"],"primary_cat":"cs.LG","authors_text":"Difan Zou, Shujie Wang, Xingjun Ma, Yunhao Chen","submitted_at":"2024-10-03T13:17:06Z","abstract_excerpt":"As diffusion probabilistic models (DPMs) become central to Generative AI (GenAI), understanding their memorization behavior is essential for evaluating risks such as data leakage, copyright infringement, and trustworthiness. While prior research finds conditional DPMs highly susceptible to data extraction attacks using explicit prompts, unconditional models are often assumed to be safe. We challenge this view by introducing \\textbf{Surrogate condItional Data Extraction (SIDE)}, a general framework that constructs data-driven surrogate conditions to enable targeted extraction from any DPM. Thro"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.02467","kind":"arxiv","version":7},"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/2410.02467/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":"2410.02467","created_at":"2026-07-05T11:46:36.370940+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.02467v7","created_at":"2026-07-05T11:46:36.370940+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.02467","created_at":"2026-07-05T11:46:36.370940+00:00"},{"alias_kind":"pith_short_12","alias_value":"NFIXUZEJOBK6","created_at":"2026-07-05T11:46:36.370940+00:00"},{"alias_kind":"pith_short_16","alias_value":"NFIXUZEJOBK6HGSP","created_at":"2026-07-05T11:46:36.370940+00:00"},{"alias_kind":"pith_short_8","alias_value":"NFIXUZEJ","created_at":"2026-07-05T11:46:36.370940+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2508.00756","citing_title":"LeakyCLIP: Extracting Training Data from CLIP","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2511.12710","citing_title":"Evolve the Method, Not the Prompts: Evolutionary Synthesis of Jailbreak Attacks on LLMs","ref_index":8,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NFIXUZEJOBK6HGSPYHMOEQY5Z6","json":"https://pith.science/pith/NFIXUZEJOBK6HGSPYHMOEQY5Z6.json","graph_json":"https://pith.science/api/pith-number/NFIXUZEJOBK6HGSPYHMOEQY5Z6/graph.json","events_json":"https://pith.science/api/pith-number/NFIXUZEJOBK6HGSPYHMOEQY5Z6/events.json","paper":"https://pith.science/paper/NFIXUZEJ"},"agent_actions":{"view_html":"https://pith.science/pith/NFIXUZEJOBK6HGSPYHMOEQY5Z6","download_json":"https://pith.science/pith/NFIXUZEJOBK6HGSPYHMOEQY5Z6.json","view_paper":"https://pith.science/paper/NFIXUZEJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.02467&json=true","fetch_graph":"https://pith.science/api/pith-number/NFIXUZEJOBK6HGSPYHMOEQY5Z6/graph.json","fetch_events":"https://pith.science/api/pith-number/NFIXUZEJOBK6HGSPYHMOEQY5Z6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NFIXUZEJOBK6HGSPYHMOEQY5Z6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NFIXUZEJOBK6HGSPYHMOEQY5Z6/action/storage_attestation","attest_author":"https://pith.science/pith/NFIXUZEJOBK6HGSPYHMOEQY5Z6/action/author_attestation","sign_citation":"https://pith.science/pith/NFIXUZEJOBK6HGSPYHMOEQY5Z6/action/citation_signature","submit_replication":"https://pith.science/pith/NFIXUZEJOBK6HGSPYHMOEQY5Z6/action/replication_record"}},"created_at":"2026-07-05T11:46:36.370940+00:00","updated_at":"2026-07-05T11:46:36.370940+00:00"}