{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:VTJDPJWI45T3TQUCVPXMMDPPUK","short_pith_number":"pith:VTJDPJWI","schema_version":"1.0","canonical_sha256":"acd237a6c8e767b9c282abeec60defa292384676b0367231376d942b869e97c3","source":{"kind":"arxiv","id":"2410.13808","version":2},"attestation_state":"computed","paper":{"title":"De-mark: Watermark Removal in Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Heng Huang, Junfeng Guo, Ruibo Chen, Yihan Wu","submitted_at":"2024-10-17T17:42:10Z","abstract_excerpt":"Watermarking techniques offer a promising way to identify machine-generated content via embedding covert information into the contents generated from language models (LMs). However, the robustness of the watermarking schemes has not been well explored. In this paper, we present De-mark, an advanced framework designed to remove n-gram-based watermarks effectively. Our method utilizes a novel querying strategy, termed random selection probing, which aids in assessing the strength of the watermark and identifying the red-green list within the n-gram watermark. Experiments on popular LMs, such as "},"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":"2410.13808","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-10-17T17:42:10Z","cross_cats_sorted":[],"title_canon_sha256":"8da71621c155a511b6ca574c5015b2498f9252ab924bdd538b85ddd2e5933e0a","abstract_canon_sha256":"0eb8f2a57452d95e769b07974fd1f29f734004454ed7c084ea3097c57783190f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:31:02.772729Z","signature_b64":"qmbH3VPb51SgrItxOfLtJ55C/WlUw+UPN2pFM8hMmmMqCNlMl2cS2hzyC++WALGt+eSFWm8yzo2FMCCe1LHLDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"acd237a6c8e767b9c282abeec60defa292384676b0367231376d942b869e97c3","last_reissued_at":"2026-07-05T11:31:02.772219Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:31:02.772219Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"De-mark: Watermark Removal in Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Heng Huang, Junfeng Guo, Ruibo Chen, Yihan Wu","submitted_at":"2024-10-17T17:42:10Z","abstract_excerpt":"Watermarking techniques offer a promising way to identify machine-generated content via embedding covert information into the contents generated from language models (LMs). However, the robustness of the watermarking schemes has not been well explored. In this paper, we present De-mark, an advanced framework designed to remove n-gram-based watermarks effectively. Our method utilizes a novel querying strategy, termed random selection probing, which aids in assessing the strength of the watermark and identifying the red-green list within the n-gram watermark. Experiments on popular LMs, such as "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.13808","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/2410.13808/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":"2410.13808","created_at":"2026-07-05T11:31:02.772278+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.13808v2","created_at":"2026-07-05T11:31:02.772278+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.13808","created_at":"2026-07-05T11:31:02.772278+00:00"},{"alias_kind":"pith_short_12","alias_value":"VTJDPJWI45T3","created_at":"2026-07-05T11:31:02.772278+00:00"},{"alias_kind":"pith_short_16","alias_value":"VTJDPJWI45T3TQUC","created_at":"2026-07-05T11:31:02.772278+00:00"},{"alias_kind":"pith_short_8","alias_value":"VTJDPJWI","created_at":"2026-07-05T11:31:02.772278+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.16377","citing_title":"GoCoMA: Hyperbolic Multimodal Representation Fusion for Large Language Model-Generated Code Attribution","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2604.11546","citing_title":"RLSpoofer: A Lightweight Evaluator for LLM Watermark Spoofing Resilience","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2604.06662","citing_title":"Towards Robust Content Watermarking Against Removal and Forgery Attacks","ref_index":5,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VTJDPJWI45T3TQUCVPXMMDPPUK","json":"https://pith.science/pith/VTJDPJWI45T3TQUCVPXMMDPPUK.json","graph_json":"https://pith.science/api/pith-number/VTJDPJWI45T3TQUCVPXMMDPPUK/graph.json","events_json":"https://pith.science/api/pith-number/VTJDPJWI45T3TQUCVPXMMDPPUK/events.json","paper":"https://pith.science/paper/VTJDPJWI"},"agent_actions":{"view_html":"https://pith.science/pith/VTJDPJWI45T3TQUCVPXMMDPPUK","download_json":"https://pith.science/pith/VTJDPJWI45T3TQUCVPXMMDPPUK.json","view_paper":"https://pith.science/paper/VTJDPJWI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.13808&json=true","fetch_graph":"https://pith.science/api/pith-number/VTJDPJWI45T3TQUCVPXMMDPPUK/graph.json","fetch_events":"https://pith.science/api/pith-number/VTJDPJWI45T3TQUCVPXMMDPPUK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VTJDPJWI45T3TQUCVPXMMDPPUK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VTJDPJWI45T3TQUCVPXMMDPPUK/action/storage_attestation","attest_author":"https://pith.science/pith/VTJDPJWI45T3TQUCVPXMMDPPUK/action/author_attestation","sign_citation":"https://pith.science/pith/VTJDPJWI45T3TQUCVPXMMDPPUK/action/citation_signature","submit_replication":"https://pith.science/pith/VTJDPJWI45T3TQUCVPXMMDPPUK/action/replication_record"}},"created_at":"2026-07-05T11:31:02.772278+00:00","updated_at":"2026-07-05T11:31:02.772278+00:00"}