{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:NGEPLVGUU7AZJB6DMCLM4WEKQK","short_pith_number":"pith:NGEPLVGU","schema_version":"1.0","canonical_sha256":"6988f5d4d4a7c19487c36096ce588a82bb379cfd4b474f41f5db14013312c674","source":{"kind":"arxiv","id":"2406.02603","version":1},"attestation_state":"computed","paper":{"title":"Distortion-free Watermarks are not Truly Distortion-free under Watermark Key Collisions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CR","authors_text":"Heng Huang, Hongyang Zhang, Junfeng Guo, Ruibo Chen, Yanshuo Chen, Yihan Wu, Zhengmian Hu","submitted_at":"2024-06-02T04:07:32Z","abstract_excerpt":"Language model (LM) watermarking techniques inject a statistical signal into LM-generated content by substituting the random sampling process with pseudo-random sampling, using watermark keys as the random seed. Among these statistical watermarking approaches, distortion-free watermarks are particularly crucial because they embed watermarks into LM-generated content without compromising generation quality. However, one notable limitation of pseudo-random sampling compared to true-random sampling is that, under the same watermark keys (i.e., key collision), the results of pseudo-random sampling"},"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":"2406.02603","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-06-02T04:07:32Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"7be6e4cc1421caf8ba187207ccdb2d25747ed12f8122b58440e3f7c8142c0a31","abstract_canon_sha256":"4b3b5ccf06a07883fd77d6f814c0a27f14c70266077952d9401a27bbc0a16c8b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:27:27.787994Z","signature_b64":"CkjKw5c4DnLl2nBF2xaJlJvK8Ja57Ogb+sGggg+5Oy3sIv1pF/wLiTP4MDc/s17nWlSnamJV3hoco8QjpCSUAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6988f5d4d4a7c19487c36096ce588a82bb379cfd4b474f41f5db14013312c674","last_reissued_at":"2026-07-05T08:27:27.787518Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:27:27.787518Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Distortion-free Watermarks are not Truly Distortion-free under Watermark Key Collisions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CR","authors_text":"Heng Huang, Hongyang Zhang, Junfeng Guo, Ruibo Chen, Yanshuo Chen, Yihan Wu, Zhengmian Hu","submitted_at":"2024-06-02T04:07:32Z","abstract_excerpt":"Language model (LM) watermarking techniques inject a statistical signal into LM-generated content by substituting the random sampling process with pseudo-random sampling, using watermark keys as the random seed. Among these statistical watermarking approaches, distortion-free watermarks are particularly crucial because they embed watermarks into LM-generated content without compromising generation quality. However, one notable limitation of pseudo-random sampling compared to true-random sampling is that, under the same watermark keys (i.e., key collision), the results of pseudo-random sampling"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.02603","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/2406.02603/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":"2406.02603","created_at":"2026-07-05T08:27:27.787582+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.02603v1","created_at":"2026-07-05T08:27:27.787582+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.02603","created_at":"2026-07-05T08:27:27.787582+00:00"},{"alias_kind":"pith_short_12","alias_value":"NGEPLVGUU7AZ","created_at":"2026-07-05T08:27:27.787582+00:00"},{"alias_kind":"pith_short_16","alias_value":"NGEPLVGUU7AZJB6D","created_at":"2026-07-05T08:27:27.787582+00:00"},{"alias_kind":"pith_short_8","alias_value":"NGEPLVGU","created_at":"2026-07-05T08:27:27.787582+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.31602","citing_title":"Robust Text Watermarking for Large Language Models via Dual Semantic Embeddings","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2606.31602","citing_title":"Robust Text Watermarking for Large Language Models via Dual Semantic Embeddings","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2605.25967","citing_title":"Hidden in Plain Tokens: Simply Robust, Gradient-Free Watermark for Synthetic Audio","ref_index":18,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NGEPLVGUU7AZJB6DMCLM4WEKQK","json":"https://pith.science/pith/NGEPLVGUU7AZJB6DMCLM4WEKQK.json","graph_json":"https://pith.science/api/pith-number/NGEPLVGUU7AZJB6DMCLM4WEKQK/graph.json","events_json":"https://pith.science/api/pith-number/NGEPLVGUU7AZJB6DMCLM4WEKQK/events.json","paper":"https://pith.science/paper/NGEPLVGU"},"agent_actions":{"view_html":"https://pith.science/pith/NGEPLVGUU7AZJB6DMCLM4WEKQK","download_json":"https://pith.science/pith/NGEPLVGUU7AZJB6DMCLM4WEKQK.json","view_paper":"https://pith.science/paper/NGEPLVGU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.02603&json=true","fetch_graph":"https://pith.science/api/pith-number/NGEPLVGUU7AZJB6DMCLM4WEKQK/graph.json","fetch_events":"https://pith.science/api/pith-number/NGEPLVGUU7AZJB6DMCLM4WEKQK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NGEPLVGUU7AZJB6DMCLM4WEKQK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NGEPLVGUU7AZJB6DMCLM4WEKQK/action/storage_attestation","attest_author":"https://pith.science/pith/NGEPLVGUU7AZJB6DMCLM4WEKQK/action/author_attestation","sign_citation":"https://pith.science/pith/NGEPLVGUU7AZJB6DMCLM4WEKQK/action/citation_signature","submit_replication":"https://pith.science/pith/NGEPLVGUU7AZJB6DMCLM4WEKQK/action/replication_record"}},"created_at":"2026-07-05T08:27:27.787582+00:00","updated_at":"2026-07-05T08:27:27.787582+00:00"}