{"as_of":"2026-08-17T13:49:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:3e47e03bec21c7e98643b7b44ffbbdd45d548e1c0622886f74293221a6ace084","coverage":[{"denominator":15,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":15,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-04T14:51:54.812148Z","state":"measured"},{"denominator":15,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":15,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-17T06:30:58.91139+00:00","state":"measured"},{"denominator":0,"lane":"inbound_itemization","note":"Pith citing papers itemized under the disclosed page cap.","records_observed":0,"source":"paper_references, paper_reference_links","source_observed_at":null,"state":"measured"},{"denominator":1,"lane":"external_citation_measurements","note":"A source-named dated measurement, never combined with another source.","records_observed":0,"source":"cited_works","source_observed_at":null,"state":"measured"}],"external_citation_measurements":[],"inbound":[],"links":{"evidence":"/evidence","html":"/paper/2509.23019/citation-record","integrity":"/paper/2509.23019/integrity","json":"/paper/2509.23019/citation-record.json","paper":"/paper/2509.23019"},"outbound":[{"citation":{"cited_paper":{"arxiv_id":"2407.21783","last_updated":"2024-11-23T23:27:33Z","snapshot_observed_at":"2026-08-13T17:20:44.002518Z","submitted_at":"2024-07-31T17:54:27Z","title":"The Llama 3 Herd of Models","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2407.21783","snapshot_observed_at":"2026-08-04T14:51:53.184797Z","title":"Jinlan Fu, See-Kiong Ng, Zhengbao Jiang, and Pengfei Liu","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2509.23019","last_updated":"2026-05-27T14:45:25Z","snapshot_observed_at":"2026-08-16T10:02:11.056373Z","submitted_at":"2025-09-27T00:24:57Z","title":"LLM Watermark Evasion via Bias Inversion","version":5},"reference_index":3,"source":"pdf_text","source_observed_at":"2026-08-04T14:51:53.184797Z"},"links":{"cited_paper":"/paper/2407.21783","citing_paper":"/paper/2509.23019"},"observation_digest":"sha256:6527892ac1dc850a458776798263451979d20cea2d68370a153dc9487730e8d7","observation_id":"e9fe4b6a-b7d5-40fc-9e3a-a6c3fcbaf024","resolution":{"observed_at":"2026-08-04T14:51:53.184797Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2006.03654","last_updated":"2021-10-06T21:02:00Z","snapshot_observed_at":"2026-08-16T13:36:21.119762Z","submitted_at":"2020-06-05T19:54:34Z","title":"DeBERTa: Decoding-enhanced BERT with Disentangled Attention","version":6},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2006.03654","snapshot_observed_at":"2026-08-04T14:51:53.297926Z","title":"Deberta: Decoding-enhanced bert with disentangled attention.arXiv preprint arXiv:2006.03654,","venue":null,"work_id":null,"year":2006},"citing_paper":{"arxiv_id":"2509.23019","last_updated":"2026-05-27T14:45:25Z","snapshot_observed_at":"2026-08-16T10:02:11.056373Z","submitted_at":"2025-09-27T00:24:57Z","title":"LLM Watermark Evasion via Bias Inversion","version":5},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-08-04T14:51:53.297926Z"},"links":{"cited_paper":"/paper/2006.03654","citing_paper":"/paper/2509.23019"},"observation_digest":"sha256:7d21df112d7c155b59f9ce14c40ce0e1556196046412acff699996347e6a063e","observation_id":"8369249a-878b-46c6-b757-f5c6b4120e3a","resolution":{"observed_at":"2026-08-04T14:51:53.297926Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2307.16230","last_updated":"2024-05-26T05:22:38Z","snapshot_observed_at":"2026-08-16T15:12:10.986853Z","submitted_at":"2023-07-30T13:43:27Z","title":"An Unforgeable Publicly Verifiable Watermark for Large Language Models","version":7},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2307.16230","snapshot_observed_at":"2026-08-04T14:51:53.677994Z","title":"An unforge- able publicly verifiable watermark for large language models.arXiv preprint arXiv:2307.16230, 2023a","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2509.23019","last_updated":"2026-05-27T14:45:25Z","snapshot_observed_at":"2026-08-16T10:02:11.056373Z","submitted_at":"2025-09-27T00:24:57Z","title":"LLM Watermark Evasion via Bias Inversion","version":5},"reference_index":7,"source":"pdf_text","source_observed_at":"2026-08-04T14:51:53.677994Z"},"links":{"cited_paper":"/paper/2307.16230","citing_paper":"/paper/2509.23019"},"observation_digest":"sha256:71dc01abea5bb8d8430e83b8c3a0fcf21e518fbb7fde70b2b66e68521e002694","observation_id":"1ae741b4-aafb-422b-8d3f-14e891f20305","resolution":{"observed_at":"2026-08-04T14:51:53.677994Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2303.16634","last_updated":"2023-05-23T22:12:16Z","snapshot_observed_at":"2026-08-02T04:02:36.848064Z","submitted_at":"2023-03-29T12:46:54Z","title":"G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2303.16634","snapshot_observed_at":"2026-08-04T14:51:53.866197Z","title":"G-eval: Nlg evaluation using gpt-4 with better human alignment.arXiv preprint arXiv:2303.16634, 2023b","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2509.23019","last_updated":"2026-05-27T14:45:25Z","snapshot_observed_at":"2026-08-16T10:02:11.056373Z","submitted_at":"2025-09-27T00:24:57Z","title":"LLM Watermark Evasion via Bias Inversion","version":5},"reference_index":8,"source":"pdf_text","source_observed_at":"2026-08-04T14:51:53.866197Z"},"links":{"cited_paper":"/paper/2303.16634","citing_paper":"/paper/2509.23019"},"observation_digest":"sha256:2190abcb84913dd8d2e5d90302ce95849666e47bf2e2e9ee37a4ac746defffa2","observation_id":"61669a49-041d-43c3-b63a-351ef3504b24","resolution":{"observed_at":"2026-08-04T14:51:53.866197Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2402.06196","last_updated":"2025-03-23T14:51:01Z","snapshot_observed_at":"2026-07-06T02:11:23.670680Z","submitted_at":"2024-02-09T05:37:09Z","title":"Large Language Models: A Survey","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2402.06196","snapshot_observed_at":"2026-08-04T14:51:54.040064Z","title":"Large language models: A survey.arXiv preprint arXiv:2402.06196,","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2509.23019","last_updated":"2026-05-27T14:45:25Z","snapshot_observed_at":"2026-08-16T10:02:11.056373Z","submitted_at":"2025-09-27T00:24:57Z","title":"LLM Watermark Evasion via Bias Inversion","version":5},"reference_index":9,"source":"pdf_text","source_observed_at":"2026-08-04T14:51:54.040064Z"},"links":{"cited_paper":"/paper/2402.06196","citing_paper":"/paper/2509.23019"},"observation_digest":"sha256:5543759314e8d874f1e868136de2d49196a3428605e8ebb9fa3de48f23f26d13","observation_id":"938f04ce-64f4-4e2b-93ab-619fe0545be2","resolution":{"observed_at":"2026-08-04T14:51:54.040064Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2405.10051","last_updated":"2024-10-26T05:11:11Z","snapshot_observed_at":"2026-08-16T13:52:11.349522Z","submitted_at":"2024-05-16T12:40:01Z","title":"MarkLLM: An Open-Source Toolkit for LLM Watermarking","version":6},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2405.10051","snapshot_observed_at":"2026-08-04T14:51:54.249837Z","title":"Gpt-4o: Multimodal and multilingual capabilities","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2509.23019","last_updated":"2026-05-27T14:45:25Z","snapshot_observed_at":"2026-08-16T10:02:11.056373Z","submitted_at":"2025-09-27T00:24:57Z","title":"LLM Watermark Evasion via Bias Inversion","version":5},"reference_index":10,"source":"pdf_text","source_observed_at":"2026-08-04T14:51:54.249837Z"},"links":{"cited_paper":"/paper/2405.10051","citing_paper":"/paper/2509.23019"},"observation_digest":"sha256:36e917c84f072e29ad84e9ae5bb7564ab6df1705add1a26f246fefc1ab744b1c","observation_id":"a06432c0-6f7d-454f-8296-5cbda27c4b5a","resolution":{"observed_at":"2026-08-04T14:51:54.249837Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1908.10084","last_updated":"2019-08-27T08:50:17Z","snapshot_observed_at":"2026-08-14T05:02:11.716316Z","submitted_at":"2019-08-27T08:50:17Z","title":"Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1908.10084","snapshot_observed_at":"2026-08-04T14:51:54.429799Z","title":"Sentence-bert: Sentence embeddings using siamese bert- networks.arXiv preprint arXiv:1908.10084,","venue":null,"work_id":null,"year":1908},"citing_paper":{"arxiv_id":"2509.23019","last_updated":"2026-05-27T14:45:25Z","snapshot_observed_at":"2026-08-16T10:02:11.056373Z","submitted_at":"2025-09-27T00:24:57Z","title":"LLM Watermark Evasion via Bias Inversion","version":5},"reference_index":11,"source":"pdf_text","source_observed_at":"2026-08-04T14:51:54.429799Z"},"links":{"cited_paper":"/paper/1908.10084","citing_paper":"/paper/2509.23019"},"observation_digest":"sha256:b52f762ca11734110e7e6dd838efb42d7dce87da13c509b526d4338c774e4dfc","observation_id":"97a6c5d6-8e18-464c-a93c-32fb092549e7","resolution":{"observed_at":"2026-08-04T14:51:54.429799Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2404.10552","last_updated":"2024-04-16T13:22:54Z","snapshot_observed_at":"2026-08-16T14:00:32.980657Z","submitted_at":"2024-04-16T13:22:54Z","title":"Unveiling the Misuse Potential of Base Large Language Models via In-Context Learning","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2404.10552","snapshot_observed_at":"2026-08-04T14:51:54.508102Z","title":"Lei Wang, Chen Ma, Xueyang Feng, Zeyu Zhang, Hao Yang, Jingsen Zhang, Zhiyuan Chen, Jiakai Tang, Xu Chen, Yankai Lin, et al","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2509.23019","last_updated":"2026-05-27T14:45:25Z","snapshot_observed_at":"2026-08-16T10:02:11.056373Z","submitted_at":"2025-09-27T00:24:57Z","title":"LLM Watermark Evasion via Bias Inversion","version":5},"reference_index":12,"source":"pdf_text","source_observed_at":"2026-08-04T14:51:54.508102Z"},"links":{"cited_paper":"/paper/2404.10552","citing_paper":"/paper/2509.23019"},"observation_digest":"sha256:6ace2af696c8813c673ccc18d32f12ff76cd4d5be19ccadd4109d37280ed2192","observation_id":"7becebb1-fd72-4a14-88b4-a227bcad8887","resolution":{"observed_at":"2026-08-04T14:51:54.508102Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2310.07710","last_updated":"2024-06-25T07:08:17Z","snapshot_observed_at":"2026-08-16T14:52:58.177757Z","submitted_at":"2023-10-11T17:57:35Z","title":"A Resilient and Accessible Distribution-Preserving Watermark for Large Language Models","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2310.07710","snapshot_observed_at":"2026-08-04T14:51:54.625719Z","title":"A resilient and accessible distribution-preserving watermark for large language models.arXiv preprint arXiv:2310.07710,","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2509.23019","last_updated":"2026-05-27T14:45:25Z","snapshot_observed_at":"2026-08-16T10:02:11.056373Z","submitted_at":"2025-09-27T00:24:57Z","title":"LLM Watermark Evasion via Bias Inversion","version":5},"reference_index":13,"source":"pdf_text","source_observed_at":"2026-08-04T14:51:54.625719Z"},"links":{"cited_paper":"/paper/2310.07710","citing_paper":"/paper/2509.23019"},"observation_digest":"sha256:2c04a4efe2544158253413acc57bd753d69cdafcd0abca60a8ead286c282fc10","observation_id":"c94bdecd-89f1-426a-9e19-e8abc6db28d0","resolution":{"observed_at":"2026-08-04T14:51:54.625719Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T14:51:54.713181Z","title":"I knew right away he had what we needed","venue":null,"work_id":null,"year":2008},"citing_paper":{"arxiv_id":"2509.23019","last_updated":"2026-05-27T14:45:25Z","snapshot_observed_at":"2026-08-16T10:02:11.056373Z","submitted_at":"2025-09-27T00:24:57Z","title":"LLM Watermark Evasion via Bias Inversion","version":5},"reference_index":14,"source":"pdf_text","source_observed_at":"2026-08-04T14:51:54.713181Z"},"links":{"citing_paper":"/paper/2509.23019"},"observation_digest":"sha256:ee58e24b5bf2cab0f0c1b6fc1d7a50f678eebb75f00fb67e32d3a9c3a8ebfa40","observation_id":"77621e0e-371e-421c-bb52-07b8a56ba33c","resolution":{"observed_at":"2026-08-04T14:51:54.713181Z","resolver_source":null,"status":"malformed_identifier"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T14:51:54.812148Z","title":"what happens after the end of the end of everything?","venue":null,"work_id":null,"year":1993},"citing_paper":{"arxiv_id":"2509.23019","last_updated":"2026-05-27T14:45:25Z","snapshot_observed_at":"2026-08-16T10:02:11.056373Z","submitted_at":"2025-09-27T00:24:57Z","title":"LLM Watermark Evasion via Bias Inversion","version":5},"reference_index":1967,"source":"pdf_text","source_observed_at":"2026-08-04T14:51:54.812148Z"},"links":{"citing_paper":"/paper/2509.23019"},"observation_digest":"sha256:4141dc097befdd4490e318ada1190862129a5f25ca2275e7dd21ccbd3814fbcd","observation_id":"4c7877ae-172b-4119-b65d-02a04e01f129","resolution":{"observed_at":"2026-08-04T14:51:54.812148Z","resolver_source":null,"status":"malformed_identifier"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2402.19361","last_updated":"2024-06-24T14:48:29Z","snapshot_observed_at":"2026-08-16T14:14:03.069592Z","submitted_at":"2024-02-29T17:12:39Z","title":"Watermark Stealing in Large Language Models","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2402.19361","snapshot_observed_at":"2026-08-04T14:51:53.560092Z","title":"Watermark stealing in large language models","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2509.23019","last_updated":"2026-05-27T14:45:25Z","snapshot_observed_at":"2026-08-16T10:02:11.056373Z","submitted_at":"2025-09-27T00:24:57Z","title":"LLM Watermark Evasion via Bias Inversion","version":5},"reference_index":1977,"source":"pdf_text","source_observed_at":"2026-08-04T14:51:53.560092Z"},"links":{"cited_paper":"/paper/2402.19361","citing_paper":"/paper/2509.23019"},"observation_digest":"sha256:84e05b9002774931954238e86fb858c19c64a915fd6ef470ebd38e60fa445234","observation_id":"4fda056d-a2b8-4f42-8b8e-dcc3376567d9","resolution":{"observed_at":"2026-08-04T14:51:53.560092Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2310.10669","last_updated":"2023-10-18T02:02:08Z","snapshot_observed_at":"2026-08-16T14:58:27.811000Z","submitted_at":"2023-09-22T12:46:38Z","title":"Unbiased Watermark for Large Language Models","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2310.10669","snapshot_observed_at":"2026-08-04T14:51:53.456418Z","title":"Unbi- ased watermark for large language models.arXiv preprint arXiv:2310.10669,","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2509.23019","last_updated":"2026-05-27T14:45:25Z","snapshot_observed_at":"2026-08-16T10:02:11.056373Z","submitted_at":"2025-09-27T00:24:57Z","title":"LLM Watermark Evasion via Bias Inversion","version":5},"reference_index":2020,"source":"pdf_text","source_observed_at":"2026-08-04T14:51:53.456418Z"},"links":{"cited_paper":"/paper/2310.10669","citing_paper":"/paper/2509.23019"},"observation_digest":"sha256:7144cd34b74f0406df61d11319c443f648437f5a5c287e1f56e75b3170457a95","observation_id":"77163a96-29d7-4e8c-94ae-92c5eb23b102","resolution":{"observed_at":"2026-08-04T14:51:53.456418Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2410.13808","last_updated":"2025-07-02T20:37:50Z","snapshot_observed_at":"2026-08-16T13:08:18.203230Z","submitted_at":"2024-10-17T17:42:10Z","title":"De-mark: Watermark Removal in Large Language Models","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2410.13808","snapshot_observed_at":"2026-08-04T14:51:52.937774Z","title":"Ruibo Chen, Yihan Wu, Junfeng Guo, and Heng Huang","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2509.23019","last_updated":"2026-05-27T14:45:25Z","snapshot_observed_at":"2026-08-16T10:02:11.056373Z","submitted_at":"2025-09-27T00:24:57Z","title":"LLM Watermark Evasion via Bias Inversion","version":5},"reference_index":2023,"source":"pdf_text","source_observed_at":"2026-08-04T14:51:52.937774Z"},"links":{"cited_paper":"/paper/2410.13808","citing_paper":"/paper/2509.23019"},"observation_digest":"sha256:1745f57f69d7083d38b98becace4c377aff2338ece49895b45d81af4cb9baf08","observation_id":"cbd55f5a-ba66-48a5-aeec-e533a0694848","resolution":{"observed_at":"2026-08-04T14:51:52.937774Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.05190","last_updated":"2025-05-11T14:24:22Z","snapshot_observed_at":"2026-08-16T09:49:05.729917Z","submitted_at":"2025-05-08T12:39:00Z","title":"Revealing Weaknesses in Text Watermarking Through Self-Information Rewrite Attacks","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.05190","snapshot_observed_at":"2026-08-04T14:51:53.101273Z","title":"Revealing weaknesses in text watermarking through self-information rewrite attacks.arXiv preprint arXiv:2505.05190,","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2509.23019","last_updated":"2026-05-27T14:45:25Z","snapshot_observed_at":"2026-08-16T10:02:11.056373Z","submitted_at":"2025-09-27T00:24:57Z","title":"LLM Watermark Evasion via Bias Inversion","version":5},"reference_index":2024,"source":"pdf_text","source_observed_at":"2026-08-04T14:51:53.101273Z"},"links":{"cited_paper":"/paper/2505.05190","citing_paper":"/paper/2509.23019"},"observation_digest":"sha256:83ae745f6624a3fba951ddbd5927e6367d666298dd9807b8352d3c32b58e3f1f","observation_id":"bc38bb9b-2771-4a3f-85fb-795218aeb810","resolution":{"observed_at":"2026-08-04T14:51:53.101273Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}}],"paper":{"arxiv_id":"2509.23019","last_updated":"2026-05-27T14:45:25Z","latest_version":5,"primary_category":"cs.CR","snapshot_observed_at":"2026-08-16T10:02:11.056373Z","submitted_at":"2025-09-27T00:24:57Z","title":"LLM Watermark Evasion via Bias Inversion"},"reference_resolution":{"displayed":15,"state_counts":{"malformed_identifier":2,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":13,"verified_exact":0,"verified_fuzzy":0},"total_outbound_references":15},"refusal":"A citation records a reference. It does not transfer a finding from one paper to another.","schema":"pith.paper-citation-record.v1","standing_sources":[{"observed_at":"2026-08-17T06:30:58.91139+00:00","source":"crossref"},{"observed_at":"2026-08-17T06:30:54.323127+00:00","source":"retraction_watch"}],"thesis":"As of 17 August 2026, this Paper Citation Record lists 15 of 15 outbound references and 0 inbound Pith citation observations for arXiv:2509.23019."}