{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:LCJUOARC3QPS52EDX6TS5TMZSK","short_pith_number":"pith:LCJUOARC","schema_version":"1.0","canonical_sha256":"5893470222dc1f2ee883bfa72ecd999285185e0f14fea932c19e2ab90202e9a2","source":{"kind":"arxiv","id":"2406.14517","version":2},"attestation_state":"computed","paper":{"title":"PostMark: A Robust Blackbox Watermark for Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.CR"],"primary_cat":"cs.LG","authors_text":"Amir Houmansadr, John Wieting, Kalpesh Krishna, Mohit Iyyer, Yapei Chang","submitted_at":"2024-06-20T17:27:14Z","abstract_excerpt":"The most effective techniques to detect LLM-generated text rely on inserting a detectable signature -- or watermark -- during the model's decoding process. Most existing watermarking methods require access to the underlying LLM's logits, which LLM API providers are loath to share due to fears of model distillation. As such, these watermarks must be implemented independently by each LLM provider. In this paper, we develop PostMark, a modular post-hoc watermarking procedure in which an input-dependent set of words (determined via a semantic embedding) is inserted into the text after the decoding"},"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":"2406.14517","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-06-20T17:27:14Z","cross_cats_sorted":["cs.AI","cs.CL","cs.CR"],"title_canon_sha256":"82c0b5605653787200a188dd6824af0d21be5e3cbb10076234970b836ee2b522","abstract_canon_sha256":"a658cfed10b2fabdd0707fa2402d381fe9b83787d5c52c16306c15eec2878adc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:19:10.623358Z","signature_b64":"qqavEfMAffwerpJWZDip7ggL+CvZ1HFutqsi8ofyZLkxKnoThP05mWJ8780C/MoJXsP42jLjpSIwH9MMZaYaCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5893470222dc1f2ee883bfa72ecd999285185e0f14fea932c19e2ab90202e9a2","last_reissued_at":"2026-07-05T09:19:10.622709Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:19:10.622709Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PostMark: A Robust Blackbox Watermark for Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.CR"],"primary_cat":"cs.LG","authors_text":"Amir Houmansadr, John Wieting, Kalpesh Krishna, Mohit Iyyer, Yapei Chang","submitted_at":"2024-06-20T17:27:14Z","abstract_excerpt":"The most effective techniques to detect LLM-generated text rely on inserting a detectable signature -- or watermark -- during the model's decoding process. Most existing watermarking methods require access to the underlying LLM's logits, which LLM API providers are loath to share due to fears of model distillation. As such, these watermarks must be implemented independently by each LLM provider. In this paper, we develop PostMark, a modular post-hoc watermarking procedure in which an input-dependent set of words (determined via a semantic embedding) is inserted into the text after the decoding"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.14517","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/2406.14517/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":"2406.14517","created_at":"2026-07-05T09:19:10.622783+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.14517v2","created_at":"2026-07-05T09:19:10.622783+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.14517","created_at":"2026-07-05T09:19:10.622783+00:00"},{"alias_kind":"pith_short_12","alias_value":"LCJUOARC3QPS","created_at":"2026-07-05T09:19:10.622783+00:00"},{"alias_kind":"pith_short_16","alias_value":"LCJUOARC3QPS52ED","created_at":"2026-07-05T09:19:10.622783+00:00"},{"alias_kind":"pith_short_8","alias_value":"LCJUOARC","created_at":"2026-07-05T09:19:10.622783+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2510.18333","citing_title":"Position: LLM Watermarking Should Align Stakeholders' Incentives for Practical Adoption","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08964","citing_title":"Trustworthy AI: Ensuring Reliability and Accountability from Models to Agents","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05443","citing_title":"SLAM: Structural Linguistic Activation Marking for Language Models","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04305","citing_title":"SWAN: Semantic Watermarking with Abstract Meaning Representation","ref_index":41,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LCJUOARC3QPS52EDX6TS5TMZSK","json":"https://pith.science/pith/LCJUOARC3QPS52EDX6TS5TMZSK.json","graph_json":"https://pith.science/api/pith-number/LCJUOARC3QPS52EDX6TS5TMZSK/graph.json","events_json":"https://pith.science/api/pith-number/LCJUOARC3QPS52EDX6TS5TMZSK/events.json","paper":"https://pith.science/paper/LCJUOARC"},"agent_actions":{"view_html":"https://pith.science/pith/LCJUOARC3QPS52EDX6TS5TMZSK","download_json":"https://pith.science/pith/LCJUOARC3QPS52EDX6TS5TMZSK.json","view_paper":"https://pith.science/paper/LCJUOARC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.14517&json=true","fetch_graph":"https://pith.science/api/pith-number/LCJUOARC3QPS52EDX6TS5TMZSK/graph.json","fetch_events":"https://pith.science/api/pith-number/LCJUOARC3QPS52EDX6TS5TMZSK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LCJUOARC3QPS52EDX6TS5TMZSK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LCJUOARC3QPS52EDX6TS5TMZSK/action/storage_attestation","attest_author":"https://pith.science/pith/LCJUOARC3QPS52EDX6TS5TMZSK/action/author_attestation","sign_citation":"https://pith.science/pith/LCJUOARC3QPS52EDX6TS5TMZSK/action/citation_signature","submit_replication":"https://pith.science/pith/LCJUOARC3QPS52EDX6TS5TMZSK/action/replication_record"}},"created_at":"2026-07-05T09:19:10.622783+00:00","updated_at":"2026-07-05T09:19:10.622783+00:00"}