{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:MPDQTYDBK4UUB73HZCQ42UVR3I","short_pith_number":"pith:MPDQTYDB","schema_version":"1.0","canonical_sha256":"63c709e061572940ff67c8a1cd52b1da30858d1a0303c33053e315958b2a44c2","source":{"kind":"arxiv","id":"2412.03123","version":2},"attestation_state":"computed","paper":{"title":"Robust Multi-bit Text Watermark with LLM-based Paraphrasers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Hang Li, Jinghan Jia, Xiaojun Xu, Yang Liu, Yuanshun Yao","submitted_at":"2024-12-04T08:43:12Z","abstract_excerpt":"We propose an imperceptible multi-bit text watermark embedded by paraphrasing with LLMs. We fine-tune a pair of LLM paraphrasers that are designed to behave differently so that their paraphrasing difference reflected in the text semantics can be identified by a trained decoder. To embed our multi-bit watermark, we use two paraphrasers alternatively to encode the pre-defined binary code at the sentence level. Then we use a text classifier as the decoder to decode each bit of the watermark. Through extensive experiments, we show that our watermarks can achieve over 99.99\\% detection AUC with sma"},"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":"2412.03123","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2024-12-04T08:43:12Z","cross_cats_sorted":[],"title_canon_sha256":"c7f02088ada24cea2f7a0d31120b5e98d70616a100df77c92f425a464f7d5e07","abstract_canon_sha256":"5d7ed2e0be36f504434865408ab3648fa9b9c5ccdbe41eff2caef0ed62582136"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:22:34.409607Z","signature_b64":"l2ntH/JvYJCNhT88qlvtNecMyMw2QY0DnG2Nw0HIpfUxfssGwopWYXZlMU4Yed42aZPicCJynsA1ashuMucvDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"63c709e061572940ff67c8a1cd52b1da30858d1a0303c33053e315958b2a44c2","last_reissued_at":"2026-07-05T11:22:34.409103Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:22:34.409103Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Robust Multi-bit Text Watermark with LLM-based Paraphrasers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Hang Li, Jinghan Jia, Xiaojun Xu, Yang Liu, Yuanshun Yao","submitted_at":"2024-12-04T08:43:12Z","abstract_excerpt":"We propose an imperceptible multi-bit text watermark embedded by paraphrasing with LLMs. We fine-tune a pair of LLM paraphrasers that are designed to behave differently so that their paraphrasing difference reflected in the text semantics can be identified by a trained decoder. To embed our multi-bit watermark, we use two paraphrasers alternatively to encode the pre-defined binary code at the sentence level. Then we use a text classifier as the decoder to decode each bit of the watermark. Through extensive experiments, we show that our watermarks can achieve over 99.99\\% detection AUC with sma"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.03123","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/2412.03123/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":"2412.03123","created_at":"2026-07-05T11:22:34.409162+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.03123v2","created_at":"2026-07-05T11:22:34.409162+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.03123","created_at":"2026-07-05T11:22:34.409162+00:00"},{"alias_kind":"pith_short_12","alias_value":"MPDQTYDBK4UU","created_at":"2026-07-05T11:22:34.409162+00:00"},{"alias_kind":"pith_short_16","alias_value":"MPDQTYDBK4UUB73H","created_at":"2026-07-05T11:22:34.409162+00:00"},{"alias_kind":"pith_short_8","alias_value":"MPDQTYDB","created_at":"2026-07-05T11:22:34.409162+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.25796","citing_title":"SAMark: A Self-Anchored Text Watermarking with Paragraph-Level Paraphrase Robustness","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12456","citing_title":"TextSeal: A Localized LLM Watermark for Provenance & Distillation Protection","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2508.11548","citing_title":"Copyright Protection for Large Language Models: A Survey of Methods, Challenges, and Trends","ref_index":158,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12456","citing_title":"TextSeal: A Localized LLM Watermark for Provenance & Distillation Protection","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08826","citing_title":"Fundamental Trade-Offs in Multi-Bit Watermarking of Stochastic Processes","ref_index":51,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MPDQTYDBK4UUB73HZCQ42UVR3I","json":"https://pith.science/pith/MPDQTYDBK4UUB73HZCQ42UVR3I.json","graph_json":"https://pith.science/api/pith-number/MPDQTYDBK4UUB73HZCQ42UVR3I/graph.json","events_json":"https://pith.science/api/pith-number/MPDQTYDBK4UUB73HZCQ42UVR3I/events.json","paper":"https://pith.science/paper/MPDQTYDB"},"agent_actions":{"view_html":"https://pith.science/pith/MPDQTYDBK4UUB73HZCQ42UVR3I","download_json":"https://pith.science/pith/MPDQTYDBK4UUB73HZCQ42UVR3I.json","view_paper":"https://pith.science/paper/MPDQTYDB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.03123&json=true","fetch_graph":"https://pith.science/api/pith-number/MPDQTYDBK4UUB73HZCQ42UVR3I/graph.json","fetch_events":"https://pith.science/api/pith-number/MPDQTYDBK4UUB73HZCQ42UVR3I/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MPDQTYDBK4UUB73HZCQ42UVR3I/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MPDQTYDBK4UUB73HZCQ42UVR3I/action/storage_attestation","attest_author":"https://pith.science/pith/MPDQTYDBK4UUB73HZCQ42UVR3I/action/author_attestation","sign_citation":"https://pith.science/pith/MPDQTYDBK4UUB73HZCQ42UVR3I/action/citation_signature","submit_replication":"https://pith.science/pith/MPDQTYDBK4UUB73HZCQ42UVR3I/action/replication_record"}},"created_at":"2026-07-05T11:22:34.409162+00:00","updated_at":"2026-07-05T11:22:34.409162+00:00"}