{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:QKTYHJ66Y2SDL4D3EFF7Y3YUCY","short_pith_number":"pith:QKTYHJ66","schema_version":"1.0","canonical_sha256":"82a783a7dec6a435f07b214bfc6f14162384a3de51be3112b12634f723ba9683","source":{"kind":"arxiv","id":"2301.10226","version":4},"attestation_state":"computed","paper":{"title":"A Watermark for Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.CR"],"primary_cat":"cs.LG","authors_text":"Ian Miers, John Kirchenbauer, Jonas Geiping, Jonathan Katz, Tom Goldstein, Yuxin Wen","submitted_at":"2023-01-24T18:52:59Z","abstract_excerpt":"Potential harms of large language models can be mitigated by watermarking model output, i.e., embedding signals into generated text that are invisible to humans but algorithmically detectable from a short span of tokens. We propose a watermarking framework for proprietary language models. The watermark can be embedded with negligible impact on text quality, and can be detected using an efficient open-source algorithm without access to the language model API or parameters. The watermark works by selecting a randomized set of \"green\" tokens before a word is generated, and then softly promoting u"},"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":"2301.10226","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-01-24T18:52:59Z","cross_cats_sorted":["cs.CL","cs.CR"],"title_canon_sha256":"4b0fcfb58eb15ed28cee65b19287161f9187f175a1302a90018d7be033c11080","abstract_canon_sha256":"c9a79609558174f591624ac440e6c7c9b0538e4f4d56bc57a9c88761d3ae14fd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:14:17.144692Z","signature_b64":"Zz/tVrq8BGFVoDyfEWKJeL2P6UIUoTdYAJ/t3QmOzKZJmKq4pil0IoyrVOUUzLOfOBhNalJAbTu104+33FYZBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"82a783a7dec6a435f07b214bfc6f14162384a3de51be3112b12634f723ba9683","last_reissued_at":"2026-07-05T08:14:17.144202Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:14:17.144202Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Watermark for Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.CR"],"primary_cat":"cs.LG","authors_text":"Ian Miers, John Kirchenbauer, Jonas Geiping, Jonathan Katz, Tom Goldstein, Yuxin Wen","submitted_at":"2023-01-24T18:52:59Z","abstract_excerpt":"Potential harms of large language models can be mitigated by watermarking model output, i.e., embedding signals into generated text that are invisible to humans but algorithmically detectable from a short span of tokens. We propose a watermarking framework for proprietary language models. The watermark can be embedded with negligible impact on text quality, and can be detected using an efficient open-source algorithm without access to the language model API or parameters. The watermark works by selecting a randomized set of \"green\" tokens before a word is generated, and then softly promoting u"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2301.10226","kind":"arxiv","version":4},"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/2301.10226/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":"2301.10226","created_at":"2026-07-05T08:14:17.144256+00:00"},{"alias_kind":"arxiv_version","alias_value":"2301.10226v4","created_at":"2026-07-05T08:14:17.144256+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2301.10226","created_at":"2026-07-05T08:14:17.144256+00:00"},{"alias_kind":"pith_short_12","alias_value":"QKTYHJ66Y2SD","created_at":"2026-07-05T08:14:17.144256+00:00"},{"alias_kind":"pith_short_16","alias_value":"QKTYHJ66Y2SDL4D3","created_at":"2026-07-05T08:14:17.144256+00:00"},{"alias_kind":"pith_short_8","alias_value":"QKTYHJ66","created_at":"2026-07-05T08:14:17.144256+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":20,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.05694","citing_title":"Beyond Heuristic Tuning: Power-Calibrated LLM Watermarking","ref_index":52,"is_internal_anchor":true},{"citing_arxiv_id":"2607.01248","citing_title":"A Practice Auditing Framework for Large Language Model Use: Collective Empiricism, Pseudo-Rational Cognition, and Governance of AI-Generated Content","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2607.00325","citing_title":"Watermarking for Proprietary Dataset Protection","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28425","citing_title":"Tool Use Enables Undetectable Steganography in Multi-Agent LLM Systems","ref_index":72,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20761","citing_title":"Findings of the Counter Turing Test: AI-Generated Text Detection","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29175","citing_title":"Direct Causation in International Humanitarian Law and the Challenge of AI-Mediated Civilian Cyber Operations","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2502.11336","citing_title":"ExaGPT: Example-Based Machine-Generated Text Detection for Human Interpretability","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12456","citing_title":"TextSeal: A Localized LLM Watermark for Provenance & Distillation Protection","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21948","citing_title":"SCI-Defense: Defending Manipulation Attacks from Generative Engine Optimization","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20761","citing_title":"Findings of the Counter Turing Test: AI-Generated Text Detection","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2303.11156","citing_title":"Can AI-Generated Text be Reliably Detected?","ref_index":64,"is_internal_anchor":false},{"citing_arxiv_id":"2605.16462","citing_title":"Asking Back: Interaction-Layer Antidistillation Watermarks","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2507.03014","citing_title":"Intrinsic Fingerprint of LLMs: Continue Training is NOT All You Need to Steal A Model!","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12456","citing_title":"TextSeal: A Localized LLM Watermark for Provenance & Distillation Protection","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2604.23338","citing_title":"A Systematic Survey of Security Threats and Defenses in LLM-Based AI Agents: A Layered Attack Surface Framework","ref_index":80,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05503","citing_title":"Chainwash: Multi-Step Rewriting Attacks on Diffusion Language Model Watermarks","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04305","citing_title":"SWAN: Semantic Watermarking with Abstract Meaning Representation","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06987","citing_title":"Response Time Enhances Alignment with Heterogeneous Preferences","ref_index":98,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07481","citing_title":"Vaporizer: Breaking Watermarking Schemes for Large Language Model Outputs","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16058","citing_title":"LLMSniffer: Detecting LLM-Generated Code via GraphCodeBERT and Supervised Contrastive Learning","ref_index":12,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QKTYHJ66Y2SDL4D3EFF7Y3YUCY","json":"https://pith.science/pith/QKTYHJ66Y2SDL4D3EFF7Y3YUCY.json","graph_json":"https://pith.science/api/pith-number/QKTYHJ66Y2SDL4D3EFF7Y3YUCY/graph.json","events_json":"https://pith.science/api/pith-number/QKTYHJ66Y2SDL4D3EFF7Y3YUCY/events.json","paper":"https://pith.science/paper/QKTYHJ66"},"agent_actions":{"view_html":"https://pith.science/pith/QKTYHJ66Y2SDL4D3EFF7Y3YUCY","download_json":"https://pith.science/pith/QKTYHJ66Y2SDL4D3EFF7Y3YUCY.json","view_paper":"https://pith.science/paper/QKTYHJ66","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2301.10226&json=true","fetch_graph":"https://pith.science/api/pith-number/QKTYHJ66Y2SDL4D3EFF7Y3YUCY/graph.json","fetch_events":"https://pith.science/api/pith-number/QKTYHJ66Y2SDL4D3EFF7Y3YUCY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QKTYHJ66Y2SDL4D3EFF7Y3YUCY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QKTYHJ66Y2SDL4D3EFF7Y3YUCY/action/storage_attestation","attest_author":"https://pith.science/pith/QKTYHJ66Y2SDL4D3EFF7Y3YUCY/action/author_attestation","sign_citation":"https://pith.science/pith/QKTYHJ66Y2SDL4D3EFF7Y3YUCY/action/citation_signature","submit_replication":"https://pith.science/pith/QKTYHJ66Y2SDL4D3EFF7Y3YUCY/action/replication_record"}},"created_at":"2026-07-05T08:14:17.144256+00:00","updated_at":"2026-07-05T08:14:17.144256+00:00"}