{"as_of":"2026-08-19T20:07:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:300ab069026f1c196c2ef677f85097fd4fa611a1654b78b8ecd995e3facc40ec","coverage":[{"denominator":30,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":30,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-15T14:19:02.340619Z","state":"measured"},{"denominator":30,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":30,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-19T06:32:44.657259+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/2608.11348/citation-record","integrity":"/paper/2608.11348/integrity","json":"/paper/2608.11348/citation-record.json","paper":"/paper/2608.11348"},"outbound":[{"citation":{"cited_paper":{"arxiv_id":"2303.08774","last_updated":"2024-03-04T06:01:33Z","snapshot_observed_at":"2026-08-17T09:58:46.058102Z","submitted_at":"2023-03-15T17:15:04Z","title":"GPT-4 Technical Report","version":6},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2303.08774","snapshot_observed_at":"2026-08-15T14:19:02.223788Z","title":"Gpt-4 technical report,","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.223788Z"},"links":{"cited_paper":"/paper/2303.08774","citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:a2e4917a1703cb94271b4a527ab9f82018071585b22247d8f58392322fecc0bb","observation_id":"892d3a04-cb5e-46a4-9039-8d0e4fa2cc56","resolution":{"observed_at":"2026-08-15T14:19:02.223788Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2307.09288","last_updated":"2023-07-19T17:08:59Z","snapshot_observed_at":"2026-08-07T12:56:43.323460Z","submitted_at":"2023-07-18T14:31:57Z","title":"Llama 2: Open Foundation and Fine-Tuned Chat Models","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2307.09288","snapshot_observed_at":"2026-08-15T14:19:02.228749Z","title":"Llama 2: Open foundation and fine-tuned chat models,","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.228749Z"},"links":{"cited_paper":"/paper/2307.09288","citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:1fa9443f33e069c07c6190bb40690f3757d17bca8d775924fe79df989099b112","observation_id":"2b70edc4-ec9a-4cdd-a398-daf5df57626e","resolution":{"observed_at":"2026-08-15T14:19:02.228749Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1708.06733","last_updated":"2019-03-11T20:45:33Z","snapshot_observed_at":"2026-08-12T15:43:59.037380Z","submitted_at":"2017-08-22T17:31:54Z","title":"BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chain","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1708.06733","snapshot_observed_at":"2026-08-15T14:19:02.233020Z","title":"Badnets: Identifying vulnera- bilities in the machine learning model supply chain,","venue":null,"work_id":null,"year":2017},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":3,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.233020Z"},"links":{"cited_paper":"/paper/1708.06733","citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:65d3b12eac653ead958307c5725c96fc5904c33f74cc5bc10b7d34cc46f0a140","observation_id":"283ed4b0-df58-4bae-b5e8-5313987ae7e3","resolution":{"observed_at":"2026-08-15T14:19:02.233020Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-15T14:19:02.762291Z","title":"Bait: Large language model backdoor scanning by inverting attack target,","venue":null,"work_id":"f1c11530-b4bf-48ed-ac90-9bd5a6386d9c","year":2025},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.237377Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:3501ffcd585b27b117faf446bb1f3c71c63e1bdc7e9500e6d9217e6dcd2c7eb0","observation_id":"312fbac0-ebac-46e1-8384-8457f2cad5de","resolution":{"observed_at":"2026-08-15T14:19:02.766769Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-15T14:19:02.748475Z","title":"Composite backdoor attacks against large language models,","venue":null,"work_id":"c5457893-ff90-4039-b4a7-1cbbf6ef3c26","year":2024},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":5,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.241474Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:8b367cccdcd4bf1fdb89d0e80b88a80d100026fbad23e84e22db8aead5c12aac","observation_id":"359a0288-b3e0-4acd-a2f3-3408cfb4b45f","resolution":{"observed_at":"2026-08-15T14:19:02.753445Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-15T14:19:02.735921Z","title":"Great, now write an article about that: The crescendo{Multi-Turn}{LLM}jailbreak attack,","venue":null,"work_id":"79528af9-21d6-4e5c-b915-bfe9b7773286","year":2025},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":6,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.245180Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:e0f9ad87452c6d7141714cc999f63ed32e5c8af325320871d35452fa4e25f73a","observation_id":"ef66ea80-4970-47c8-a6f0-b85d2a2ac616","resolution":{"observed_at":"2026-08-15T14:19:02.739865Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2305.17493","last_updated":"2024-04-14T05:20:10Z","snapshot_observed_at":"2026-08-16T16:09:49.526553Z","submitted_at":"2023-05-27T15:10:41Z","title":"The Curse of Recursion: Training on Generated Data Makes Models Forget","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2305.17493","snapshot_observed_at":"2026-08-15T14:19:02.249695Z","title":"The curse of recursion: Training on generated data makes models forget,","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":7,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.249695Z"},"links":{"cited_paper":"/paper/2305.17493","citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:494fe21701b9e6606fd0f7354a84b942f70bf4c7c07db570e802955b15b6fc43","observation_id":"34204acb-43cc-4cc0-b53c-8c2c78eec1b7","resolution":{"observed_at":"2026-08-15T14:19:02.249695Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-15T14:19:02.720349Z","title":"Chain-of- scrutiny: Detecting backdoor attacks for large language models,","venue":null,"work_id":"a9745db5-eb47-4101-ab1f-29456071db00","year":2025},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":8,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.253318Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:b96e3a4ba3964cfbed9c21733cc5e10afc4e8a9d82f5a321b5c1ba2d6102731a","observation_id":"65b61a9f-780b-4534-86a9-81f8793f45e3","resolution":{"observed_at":"2026-08-15T14:19:02.725146Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-15T14:19:02.705250Z","title":"When backdoors speak: Understanding llm backdoor attacks through model-generated explana- tions,","venue":null,"work_id":"79b9e4eb-a759-4e47-8bba-87f684a2b782","year":2025},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":9,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.256551Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:282bba7522f873b627d8d67d8d57970f888b42da1d1dfffcf32f4d2bff9d2fa0","observation_id":"6ef63ed5-92b5-4a82-b073-304bab61c640","resolution":{"observed_at":"2026-08-15T14:19:02.710010Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-15T14:19:02.689706Z","title":"Mitigating backdoor threats to large language models: Advancement and challenges,","venue":null,"work_id":"96b780b7-d6c6-418e-95d3-303ae0f148b8","year":2024},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":10,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.259713Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:cc3795edf127b315f03d5d4ec8dcbf04d06d88dafcd75037855088d89ebb1176","observation_id":"1783c0dc-6e15-419d-a421-b1a411efe747","resolution":{"observed_at":"2026-08-15T14:19:02.694701Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-15T14:19:02.676389Z","title":"Onion: A simple and effective defense against textual backdoor attacks,","venue":null,"work_id":"361320e0-a601-44f5-a6eb-258d3c891f72","year":2021},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":11,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.263347Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:f9027bd387a3973c041648ce38e533db430c0fdb1f4df9d4dce5c9bb19264333","observation_id":"dc60ef13-b1a4-4d35-b1bf-3c619808589c","resolution":{"observed_at":"2026-08-15T14:19:02.680072Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-15T14:19:02.664372Z","title":"Iclscan: Detecting backdoors in black-box large language models via targeted in-context illumination,","venue":null,"work_id":"1be9b2f3-3e51-4e90-a34d-518524666387","year":2026},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":12,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.266900Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:6cb561db1728863d0cd4b4d428f1668f5c169a11a36df346165058687afb6b5f","observation_id":"0ac380f3-35f9-498e-b455-5f7bab67fcad","resolution":{"observed_at":"2026-08-15T14:19:02.668404Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-15T14:19:02.652436Z","title":"Bite: Textual backdoor attacks with iterative trigger injection,","venue":null,"work_id":"ad733fdb-4660-4f47-8e57-11d41061474d","year":2023},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":13,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.271441Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:893d610562dda905383db3c20c9aed441d1637688915f42ac6127cf4f050cd63","observation_id":"16bee2bd-c47c-45ba-921f-d527a150ab71","resolution":{"observed_at":"2026-08-15T14:19:02.656664Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-15T14:19:02.637328Z","title":"Imbert: Making bert immune to insertion-based backdoor attacks,","venue":null,"work_id":"5eeb81c6-7152-4b78-835b-f2eab9f4a8c3","year":2023},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":14,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.275412Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:85eed2036064e52d3be2ec6ed127fcc69600f381068b66ffee91212744cfc6db","observation_id":"18c468eb-96bf-45ed-9fba-f3f9ceae5c92","resolution":{"observed_at":"2026-08-15T14:19:02.642386Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-15T14:19:02.622288Z","title":"Granite 4.0 language models,","venue":null,"work_id":"05e15c9d-4026-4783-9db2-d730f6cfb057","year":2025},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":15,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.278723Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:b308396c0f697d2598e0c57abea4aefc684d11bd2db3162d7d6fa38ea226cadc","observation_id":"eee3c784-70a4-486b-aa10-adb48b46e1ae","resolution":{"observed_at":"2026-08-15T14:19:02.627030Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"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":"2026-08-15T14:19:02.282170Z","title":"The llama 3 herd of models,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":16,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.282170Z"},"links":{"cited_paper":"/paper/2407.21783","citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:7b4cdf68118cfe7963a0b32157509ca17e73284596d1a1354ee126c3ae4dd8b5","observation_id":"fc4ce676-9301-4d81-bce1-d61394a5a6d3","resolution":{"observed_at":"2026-08-15T14:19:02.282170Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2310.06825","last_updated":"2023-10-10T17:54:58Z","snapshot_observed_at":"2026-08-17T20:30:34.016254Z","submitted_at":"2023-10-10T17:54:58Z","title":"Mistral 7B","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2310.06825","snapshot_observed_at":"2026-08-15T14:19:02.285655Z","title":"Mistral 7b,","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":17,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.285655Z"},"links":{"cited_paper":"/paper/2310.06825","citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:de282f74947b2de69f962a38c266e6eb098568feed0c1188059ed046a211b441","observation_id":"9dc4d025-e53c-4c26-8267-71bcf7a8afac","resolution":{"observed_at":"2026-08-15T14:19:02.285655Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2505.09388","last_updated":"2025-05-14T13:41:34Z","snapshot_observed_at":"2026-07-06T02:11:23.670680Z","submitted_at":"2025-05-14T13:41:34Z","title":"Qwen3 Technical Report","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2505.09388","snapshot_observed_at":"2026-08-15T14:19:02.289933Z","title":"Qwen3 technical report,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":18,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.289933Z"},"links":{"cited_paper":"/paper/2505.09388","citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:0a76a6ebcb70d67b6d5003b6c015c55c54fac35d158adf98cf7ebadc77f2da41","observation_id":"d3a0426f-187e-4c2f-8bb9-92e0c8290624","resolution":{"observed_at":"2026-08-15T14:19:02.289933Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2503.19786","last_updated":"2025-03-25T15:52:34Z","snapshot_observed_at":"2026-07-06T02:11:23.670680Z","submitted_at":"2025-03-25T15:52:34Z","title":"Gemma 3 Technical Report","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2503.19786","snapshot_observed_at":"2026-08-15T14:19:02.293255Z","title":"Gemma 3 technical report,","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":19,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.293255Z"},"links":{"cited_paper":"/paper/2503.19786","citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:d50948a0c94254d470c8beada3ef770db29d1c03ff6ea2616ef38c275adab822","observation_id":"0fa77aba-a1ca-4862-9477-484b689c209b","resolution":{"observed_at":"2026-08-15T14:19:02.293255Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2412.08905","last_updated":"2024-12-12T03:37:41Z","snapshot_observed_at":"2026-08-14T03:45:52.225630Z","submitted_at":"2024-12-12T03:37:41Z","title":"Phi-4 Technical Report","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2412.08905","snapshot_observed_at":"2026-08-15T14:19:02.297588Z","title":"Phi- 4 technical report,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":20,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.297588Z"},"links":{"cited_paper":"/paper/2412.08905","citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:68abe9e29738dbfa5619944a7f35765bcec72c6745201fa8bcc3af49f84de885","observation_id":"8bdef80a-68b4-40a0-a1ec-3ce828c604ec","resolution":{"observed_at":"2026-08-15T14:19:02.297588Z","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":"2026-08-15T14:19:02.301879Z","title":"Lora: Low-rank adaptation of large language models","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":21,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.301879Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:4a9b8e77e189fca4b000e5a62875832f647307b396a68c207358d83efdafe409","observation_id":"20071e8f-7a1e-4828-a557-82a136333501","resolution":{"observed_at":"2026-08-15T14:19:02.301879Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2403.13372","last_updated":"2024-06-27T22:44:48Z","snapshot_observed_at":"2026-08-17T02:51:06.474773Z","submitted_at":"2024-03-20T08:08:54Z","title":"LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Models","version":4},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2403.13372","snapshot_observed_at":"2026-08-15T14:19:02.306366Z","title":"Llamafactory: Unified efficient fine-tuning of 100+ language models,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":22,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.306366Z"},"links":{"cited_paper":"/paper/2403.13372","citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:18fba4ed848b8038e059ba2e7c9f00b319c04cc9b80afab8fce4a4c94640c37a","observation_id":"24df339c-ab20-4d97-a4b3-6b58b89040f5","resolution":{"observed_at":"2026-08-15T14:19:02.306366Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-15T14:19:02.594999Z","title":"Axolotl: Open source llm post- training,","venue":null,"work_id":"c0dc97e6-bfc4-4d95-9d93-ab8f1bdb8e16","year":2023},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":23,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.310545Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:d23a58fc5728455d1f4d72d7ffc4da4467a8ce326889218e3d7a3fc7e7ad7d07","observation_id":"1babca80-b9db-409d-a3fa-2fff6c9ba181","resolution":{"observed_at":"2026-08-15T14:19:02.600030Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2305.14314","last_updated":"2023-05-23T17:50:33Z","snapshot_observed_at":"2026-08-13T23:56:42.354755Z","submitted_at":"2023-05-23T17:50:33Z","title":"QLoRA: Efficient Finetuning of Quantized LLMs","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2305.14314","snapshot_observed_at":"2026-08-15T14:19:02.314472Z","title":"Qlora: Efficient finetuning of quantized llms, 2023,","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":24,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.314472Z"},"links":{"cited_paper":"/paper/2305.14314","citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:80d00972a264b058b17fa734e03fe35ebdb55b7af43cad8d56e2c54c184ee5bc","observation_id":"12889811-b456-485a-ae87-767cf440091c","resolution":{"observed_at":"2026-08-15T14:19:02.314472Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-15T14:19:02.582476Z","title":"Unsloth,","venue":null,"work_id":"ede969d1-bf18-4409-a895-121b05a37072","year":2023},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":25,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.318786Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:f8810f65e4a15a4e64c1d176f47f0ab1dd59d55897f5f02cf8b823b7dbfe90a9","observation_id":"0cb6a8d7-c3cd-4e6a-8bc5-0abe374b30a1","resolution":{"observed_at":"2026-08-15T14:19:02.586525Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-15T14:19:02.570496Z","title":"Cleangen: Mitigating backdoor attacks for generation tasks in large language models,","venue":null,"work_id":"6e584b1c-2a2b-46b9-9e95-c912336012b9","year":2024},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":26,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.322825Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:890fffbdab13639c2f149749a64105358c05ab17bcd3dbe3fa759f391df5bb46","observation_id":"516006b6-f8b1-4f91-8bc4-82fd7408f646","resolution":{"observed_at":"2026-08-15T14:19:02.574543Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2411.12768","last_updated":"2025-06-11T12:40:14Z","snapshot_observed_at":"2026-08-18T17:46:12.528256Z","submitted_at":"2024-11-18T07:52:12Z","title":"CROW: Eliminating Backdoors from Large Language Models via Internal Consistency Regularization","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2411.12768","snapshot_observed_at":"2026-08-15T14:19:02.327400Z","title":"Crow: Eliminating backdoors from large language models via internal consistency regularization,","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":27,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.327400Z"},"links":{"cited_paper":"/paper/2411.12768","citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:e6e056a32ce7766a626b4656f01df9928326ad03ed434c10b0ed1428607a52b7","observation_id":"9613aadd-a1ce-48ab-9e68-3b326bb32422","resolution":{"observed_at":"2026-08-15T14:19:02.327400Z","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":"2026-08-15T14:19:02.331599Z","title":"Fine-pruning: Defending against backdooring attacks on deep neural networks,","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":28,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.331599Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:db2f9736b5714c4450c57a65cef491605cb48d1abdd59436d619a7be52ff25d9","observation_id":"223df19d-df49-4f60-aac6-128b7205f104","resolution":{"observed_at":"2026-08-15T14:19:02.331599Z","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":"2026-08-15T14:19:02.336073Z","title":"Neural cleanse: Identifying and mitigating backdoor attacks in neural networks,","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":29,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.336073Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:4f11334e60e1fcb293d10c101bd5546cb1a25c3049f51e852fc1ef30da3b6cad","observation_id":"a91ea1b2-a459-419b-8c48-d80b0bd788f7","resolution":{"observed_at":"2026-08-15T14:19:02.336073Z","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":"raw_reference","pith_arxiv_id":null,"snapshot_observed_at":"2026-08-15T14:19:02.539612Z","title":"Test-time backdoor mitigation for black-box large language models with defensive demonstrations,","venue":null,"work_id":"ab78ee67-29c1-49fe-9380-35d75c290f4d","year":2025},"citing_paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs","version":1},"reference_index":30,"source":"pdf_text","source_observed_at":"2026-08-15T14:19:02.340619Z"},"links":{"citing_paper":"/paper/2608.11348"},"observation_digest":"sha256:f5055400adfb6330d994662fd9ad220550b3f0ddd3eb51f8967d3a23341ab981","observation_id":"5473567d-a513-432c-aa58-2ff1ae800009","resolution":{"observed_at":"2026-08-15T14:19:02.546015Z","resolver_source":"raw_fallback","status":"verified_fuzzy"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-19T06:32:44.657259+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"state":"measured"}}],"paper":{"arxiv_id":"2608.11348","last_updated":"2026-08-11T18:53:49Z","latest_version":1,"primary_category":"cs.CR","snapshot_observed_at":"2026-08-19T16:23:07.701380Z","submitted_at":"2026-08-11T18:53:49Z","title":"An Empirical Study of Output-to-Input Loops for Black-Box Backdoor Detection in Fine-Tuned Open-Weight LLMs"},"reference_resolution":{"displayed":30,"state_counts":{"malformed_identifier":0,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":15,"verified_exact":0,"verified_fuzzy":15},"total_outbound_references":30},"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":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"thesis":"As of 19 August 2026, this Paper Citation Record lists 30 of 30 outbound references and 0 inbound Pith citation observations for arXiv:2608.11348."}