{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ATUZ273TCVWUAROS4UGUJVDXNB","short_pith_number":"pith:ATUZ273T","schema_version":"1.0","canonical_sha256":"04e99d7f73156d4045d2e50d44d4776840d933709a7d3e0131c76c22d2d06c0c","source":{"kind":"arxiv","id":"2504.20111","version":1},"attestation_state":"computed","paper":{"title":"Forging and Removing Latent-Noise Diffusion Watermarks Using a Single Image","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Anubhav Jain, Julian Togelius, Naoki Murata, Nasir Memon, Niv Cohen, Takashi Shibuya, Yuhta Takida, Yuki Mitsufuji, Yuya Kobayashi","submitted_at":"2025-04-27T18:29:26Z","abstract_excerpt":"Watermarking techniques are vital for protecting intellectual property and preventing fraudulent use of media. Most previous watermarking schemes designed for diffusion models embed a secret key in the initial noise. The resulting pattern is often considered hard to remove and forge into unrelated images. In this paper, we propose a black-box adversarial attack without presuming access to the diffusion model weights. Our attack uses only a single watermarked example and is based on a simple observation: there is a many-to-one mapping between images and initial noises. There are regions in 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":"2504.20111","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-04-27T18:29:26Z","cross_cats_sorted":[],"title_canon_sha256":"65805d95b1b2e2ad1da3b8e78f9b49c9f187c3aa4c78a93ba5233802118a54d3","abstract_canon_sha256":"b07fa32f146e32943655ef6b10f920064da4895f17fff349a0a5c7c5cc8d4c8d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:55:36.742281Z","signature_b64":"qsjDoBF623pvNPtIEd/2+ZNgP+ZM15sFR4lMnivjKevIB0oqfBnW2fPP6S3ptfThPnuqg+Of8GTnxeewFtY6BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"04e99d7f73156d4045d2e50d44d4776840d933709a7d3e0131c76c22d2d06c0c","last_reissued_at":"2026-07-05T10:55:36.741798Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:55:36.741798Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Forging and Removing Latent-Noise Diffusion Watermarks Using a Single Image","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Anubhav Jain, Julian Togelius, Naoki Murata, Nasir Memon, Niv Cohen, Takashi Shibuya, Yuhta Takida, Yuki Mitsufuji, Yuya Kobayashi","submitted_at":"2025-04-27T18:29:26Z","abstract_excerpt":"Watermarking techniques are vital for protecting intellectual property and preventing fraudulent use of media. Most previous watermarking schemes designed for diffusion models embed a secret key in the initial noise. The resulting pattern is often considered hard to remove and forge into unrelated images. In this paper, we propose a black-box adversarial attack without presuming access to the diffusion model weights. Our attack uses only a single watermarked example and is based on a simple observation: there is a many-to-one mapping between images and initial noises. There are regions in the "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.20111","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/2504.20111/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":"2504.20111","created_at":"2026-07-05T10:55:36.741862+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.20111v1","created_at":"2026-07-05T10:55:36.741862+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.20111","created_at":"2026-07-05T10:55:36.741862+00:00"},{"alias_kind":"pith_short_12","alias_value":"ATUZ273TCVWU","created_at":"2026-07-05T10:55:36.741862+00:00"},{"alias_kind":"pith_short_16","alias_value":"ATUZ273TCVWUAROS","created_at":"2026-07-05T10:55:36.741862+00:00"},{"alias_kind":"pith_short_8","alias_value":"ATUZ273T","created_at":"2026-07-05T10:55:36.741862+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2503.12172","citing_title":"SEAL: Semantic Aware Image Watermarking","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2507.07871","citing_title":"Mitigating Watermark Forgery in Generative Models via Randomized Key Selection","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09319","citing_title":"PGID: Progressive Guided Inversion and Denoising for Robust Watermark Detection","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2604.11720","citing_title":"On the Robustness of Watermarking for Autoregressive Image Generation","ref_index":15,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ATUZ273TCVWUAROS4UGUJVDXNB","json":"https://pith.science/pith/ATUZ273TCVWUAROS4UGUJVDXNB.json","graph_json":"https://pith.science/api/pith-number/ATUZ273TCVWUAROS4UGUJVDXNB/graph.json","events_json":"https://pith.science/api/pith-number/ATUZ273TCVWUAROS4UGUJVDXNB/events.json","paper":"https://pith.science/paper/ATUZ273T"},"agent_actions":{"view_html":"https://pith.science/pith/ATUZ273TCVWUAROS4UGUJVDXNB","download_json":"https://pith.science/pith/ATUZ273TCVWUAROS4UGUJVDXNB.json","view_paper":"https://pith.science/paper/ATUZ273T","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.20111&json=true","fetch_graph":"https://pith.science/api/pith-number/ATUZ273TCVWUAROS4UGUJVDXNB/graph.json","fetch_events":"https://pith.science/api/pith-number/ATUZ273TCVWUAROS4UGUJVDXNB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ATUZ273TCVWUAROS4UGUJVDXNB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ATUZ273TCVWUAROS4UGUJVDXNB/action/storage_attestation","attest_author":"https://pith.science/pith/ATUZ273TCVWUAROS4UGUJVDXNB/action/author_attestation","sign_citation":"https://pith.science/pith/ATUZ273TCVWUAROS4UGUJVDXNB/action/citation_signature","submit_replication":"https://pith.science/pith/ATUZ273TCVWUAROS4UGUJVDXNB/action/replication_record"}},"created_at":"2026-07-05T10:55:36.741862+00:00","updated_at":"2026-07-05T10:55:36.741862+00:00"}