{"as_of":"2026-08-08T17:50:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:70da3f15e22dd85753f84250a3d7f17f2a9a9702388cb9ec6f41d79e69377ccf","coverage":[{"denominator":49,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":49,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-02T18:40:39.638638Z","state":"measured"},{"denominator":51,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":51,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-08T06:32:00.761636+00:00","state":"measured"},{"denominator":2,"lane":"inbound_itemization","note":"Pith citing papers itemized under the disclosed page cap.","records_observed":2,"source":"paper_references, paper_reference_links","source_observed_at":"2026-06-30T20:53:04.274840Z","state":"measured"},{"denominator":1,"lane":"external_citation_measurements","note":"A source-named dated measurement, never combined with another source.","records_observed":0,"source":"arxiv_reference","source_observed_at":"2026-06-30T20:55:03.970803Z","state":"measured"}],"external_citation_measurements":[],"inbound":[{"citation":{"cited_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"cited_work":{"arxiv_id":"2603.07466","doi":null,"metadata_source":"arxiv_reference","pith_arxiv_id":"2603.07466","snapshot_observed_at":"2026-07-23T02:23:27.054931Z","title":"Thomas Hou, and Wenjing Lou","venue":null,"work_id":"17f2ed4f-1878-4c90-a987-a907e9ac6f83","year":2026},"citing_paper":{"arxiv_id":"2604.18179","last_updated":"2026-05-26T01:05:19Z","snapshot_observed_at":"2026-08-02T21:42:58.009676Z","submitted_at":"2026-04-20T12:34:56Z","title":"Committed SAE-Feature Traces for Audited-Session Substitution Detection in Hosted LLMs","version":1},"reference_index":20,"source":"pdf_text","source_observed_at":"2026-05-10T04:51:36.160507Z"},"links":{"cited_paper":"/paper/2603.07466","citing_paper":"/paper/2604.18179"},"observation_digest":"sha256:9df8aebe8e3320e67acf3765fc3885e25bf2c570d94d3525f4ba256d4c7f9338","observation_id":"4842293a-c8c7-4bed-bc5b-5b94fea1796b","resolution":{"observed_at":"2026-07-23T02:23:27.054931Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}},{"citation":{"cited_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"cited_work":{"arxiv_id":"2603.07466","doi":null,"metadata_source":"arxiv_reference","pith_arxiv_id":"2603.07466","snapshot_observed_at":"2026-07-23T02:23:27.054931Z","title":"Thomas Hou, and Wenjing Lou","venue":null,"work_id":"17f2ed4f-1878-4c90-a987-a907e9ac6f83","year":2026},"citing_paper":{"arxiv_id":"2605.15164","last_updated":"2026-05-14T17:54:28Z","snapshot_observed_at":"2026-08-02T22:54:03.046548Z","submitted_at":"2026-05-14T17:54:28Z","title":"Position: Behavioural Assurance Cannot Verify the Safety Claims Governance Now Demands","version":1},"reference_index":70,"source":"pdf_text","source_observed_at":"2026-06-30T20:53:04.274840Z"},"links":{"cited_paper":"/paper/2603.07466","citing_paper":"/paper/2605.15164"},"observation_digest":"sha256:4558fdb4e4e654146583d57c50d97ff8f688157d9f7ae597f0af985a4d836c84","observation_id":"34855d9d-b935-4bac-8bc0-9b1cb87316f0","resolution":{"observed_at":"2026-07-23T02:23:27.054931Z","resolver_source":"arxiv_id","status":"verified_exact"},"standing_notice":{"events":[],"observation":"No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.","reason":null,"source_receipts":[{"observed_at":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"state":"measured"}}],"links":{"evidence":"/evidence","html":"/paper/2603.07466/citation-record","integrity":"/paper/2603.07466/integrity","json":"/paper/2603.07466/citation-record.json","paper":"/paper/2603.07466"},"outbound":[{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-02T18:40:34.727117Z","title":"https://cloud.google.com/blog/products/identity-security/how- confidential-computing-lays-the-foundation-for-trusted-ai","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:34.727117Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:967b226b9b204f5757604bad05ff190b2a824bea73d8701fd09076e9cd342c74","observation_id":"332679df-bf6e-46d9-a6b3-8c7b72e7986e","resolution":{"observed_at":"2026-08-02T18:40:34.727117Z","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":"2026-08-02T18:40:34.780988Z","title":"https://aws.amazon.com/blogs/machine-learning/large- language-model-inference-over-confidential-data-using-aws-nitro- enclaves/","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:34.780988Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:a5764f8c8a2a45fe1b0564a28e99684995cad21b12fdaf12fe0c8a1c3e011953","observation_id":"87ef9917-e7ed-4ed4-b073-dd03015fadb9","resolution":{"observed_at":"2026-08-02T18:40:34.780988Z","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":"2026-08-02T18:40:34.849268Z","title":"Zero-knowledge proofs of training for deep neural networks","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":3,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:34.849268Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:02cd2dfb666201b2bdbac35d71be8ae3b58de2780ccccfb919aa7b0106d47272","observation_id":"4116905f-c94c-4666-bb80-001f3378850d","resolution":{"observed_at":"2026-08-02T18:40:34.849268Z","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":"2026-08-02T18:40:34.919497Z","title":"Trustzone: Integrated hardware and software security.ARM White Paper(2004)","venue":null,"work_id":null,"year":2004},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:34.919497Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:5f07b9643116ce535c571a308ed4bacec6ab7cbe07f4aa88fb93b5e764512760","observation_id":"81b85c52-9c1f-4d8d-948c-72db40a89a11","resolution":{"observed_at":"2026-08-02T18:40:34.919497Z","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":"2026-08-02T18:40:35.022399Z","title":"Fine-tune claude 3 haiku in amazon bedrock, July 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":5,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:35.022399Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:229a2d11c67b322503881c0379ed0f3c772bf6b70bdce0a720d1e860e39c8061","observation_id":"efd47ac8-054f-4d91-9563-b14b9100aeb2","resolution":{"observed_at":"2026-08-02T18:40:35.022399Z","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":"2026-08-02T18:40:35.101687Z","title":"Provable data possession at untrusted stores","venue":null,"work_id":null,"year":2007},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":6,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:35.101687Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:7c2422823e19c12e7ddbc5873c52c5c6a77d237b3801f410771d38d82b5ec08c","observation_id":"4083daef-51be-4b2b-bcf9-39c76630d3be","resolution":{"observed_at":"2026-08-02T18:40:35.101687Z","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":"2026-08-02T18:40:35.163087Z","title":"E.,ANDVIJ, M","venue":null,"work_id":null,"year":2017},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":7,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:35.163087Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:24280622ae9b343f1dec84dceaffc090ca5dac305324176a51d48e4a3eaf2777","observation_id":"770e6afb-1c4b-413f-a7f6-d7bed90263b2","resolution":{"observed_at":"2026-08-02T18:40:35.163087Z","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":"2026-08-02T18:40:35.238627Z","title":"D., ZHAO, J.,ANDKOUSHANFAR, F","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":8,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:35.238627Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:cff9642ff745812af43ec75d5053c4a18fa6b7ec5308a32d92eeaaa01bebfa21","observation_id":"ab003c40-c49c-4bf1-9546-6606abffe205","resolution":{"observed_at":"2026-08-02T18:40:35.238627Z","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":"2026-08-02T18:40:35.343313Z","title":"Tune gemini models by using supervised fine-tuning, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":9,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:35.343313Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:0d85656d34ad331fa6a51e5bc187e8835a212dbdac0f0c2a014cf34a31b228d9","observation_id":"31450c30-0b24-44c0-8f3a-df4ca21e759a","resolution":{"observed_at":"2026-08-02T18:40:35.343313Z","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":"2026-08-02T18:40:35.449271Z","title":"Azure openai in azure ai foundry models, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":10,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:35.449271Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:ec91da693ee78372c1817d7447a5a7499ae9e8537b3a3fc838656607be803e0c","observation_id":"25ac99d9-592d-456e-a9f9-d03305e06b81","resolution":{"observed_at":"2026-08-02T18:40:35.449271Z","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":"2026-08-02T18:40:35.528823Z","title":"Host llms with nvidia gpus on oci, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":11,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:35.528823Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:ca64b4f249456da7edb468a116b8a54b4923462c9eb78d107234e333260129ca","observation_id":"f2117429-17e9-406e-a184-3a43c1f7e9d3","resolution":{"observed_at":"2026-08-02T18:40:35.528823Z","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":"2026-08-02T18:40:35.600123Z","title":"Intel sgx explained","venue":null,"work_id":null,"year":2016},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":12,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:35.600123Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:4ade46cfd4bf1cd33e2298205cf6a494bd504a475326c48e083847f5af3f6896","observation_id":"2e7e163a-741c-42d2-ab90-c5dd623ff9b0","resolution":{"observed_at":"2026-08-02T18:40:35.600123Z","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":"2026-08-02T18:40:35.680033Z","title":"Guardain: Protecting emerging generative ai workloads on heterogeneous npu","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":13,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:35.680033Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:90df6d0a30c9b37a76d04806ca3cff98d4c674b62c1b3d81638fc33ceb0e95c7","observation_id":"f7532a31-09cd-47e9-9c15-9fed2539ce07","resolution":{"observed_at":"2026-08-02T18:40:35.680033Z","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":"2026-08-02T18:40:35.782709Z","title":null,"venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":14,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:35.782709Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:919edc3620dbc043de9039939cdfb83bd78d79a0453c8c5131e655de4c24a851","observation_id":"78149fc9-fe09-4c8a-8062-0a21655eff0b","resolution":{"observed_at":"2026-08-02T18:40:35.782709Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2407.21783","last_updated":"2024-11-23T23:27:33Z","snapshot_observed_at":"2026-07-06T18:55:11.576666Z","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":"2026-08-02T18:40:35.874672Z","title":null,"venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":15,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:35.874672Z"},"links":{"cited_paper":"/paper/2407.21783","citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:ad509edae1d4b57e9327502a67cab0e678c152305bae83d853b9c416e8324d3e","observation_id":"f65a52bb-bcf9-4fff-acb1-58cf501df0fe","resolution":{"observed_at":"2026-08-02T18:40:35.874672Z","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":"2026-08-02T18:40:35.950506Z","title":"H., LIPTON, Z","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":16,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:35.950506Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:2445554f109611aa4cfd879221e4c6123b599bf4da76941808a2de076bae7724","observation_id":"6cfbb277-c8d2-425e-aa2d-a65852318692","resolution":{"observed_at":"2026-08-02T18:40:35.950506Z","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":"2026-08-02T18:40:36.020273Z","title":"Scalable zero-knowledge proofs for non-linear functions in machine learning","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":17,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:36.020273Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:cf9599cf30760421e32c11160e4408bb681edf7b229a2f05c9b3f7986abc6c33","observation_id":"5d59ef8b-3734-4586-850d-ee1604e89424","resolution":{"observed_at":"2026-08-02T18:40:36.020273Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2106.09685","last_updated":"2021-10-16T18:40:34Z","snapshot_observed_at":"2026-08-07T07:43:16.294957Z","submitted_at":"2021-06-17T17:37:18Z","title":"LoRA: Low-Rank Adaptation of Large Language Models","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2106.09685","snapshot_observed_at":"2026-08-02T18:40:36.098375Z","title":"J., SHEN, Y., WALLIS, P., ALLEN-ZHU, Z., LI, Y., WANG, S., ANDCHEN, W","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":18,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:36.098375Z"},"links":{"cited_paper":"/paper/2106.09685","citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:1c8c51c5baee67bbbce7039060ffe9fb486f840afe59f9c02dd14284c5811ab8","observation_id":"4c948981-ce53-4aaa-b17e-4ec9e0ad4c49","resolution":{"observed_at":"2026-08-02T18:40:36.098375Z","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":"2026-08-02T18:40:36.165450Z","title":"A survey on hardware accelerators for large language models.Applied Sciences 15, 2 (Jan","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":19,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:36.165450Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:1c90491c938fbfa60470561023bded07900980a79c1290631cab7ca0e27231e6","observation_id":"16b584a9-b3f4-484d-a942-734c239b64bc","resolution":{"observed_at":"2026-08-02T18:40:36.165450Z","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":"2026-08-02T18:40:36.235754Z","title":"Gramine-tdx: A lightweight os kernel for confidential vms","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":20,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:36.235754Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:b96434dec1ebaf88be2d6da6cf52c5c8500ed7148a108e68a87b64444a44b889","observation_id":"60510822-eae2-4e21-993c-139366e97c37","resolution":{"observed_at":"2026-08-02T18:40:36.235754Z","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":"2026-08-02T18:40:36.309762Z","title":"Keystone: An open framework for architecting trusted execution environments","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":21,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:36.309762Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:d8ee28f2b4b744033556502560474eeb30e4ea03326f32e97a598f1b57a77128","observation_id":"cf9bb2b3-b9aa-4dac-b74e-6941ccb48119","resolution":{"observed_at":"2026-08-02T18:40:36.309762Z","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":"2026-08-02T18:40:36.423186Z","title":"Character- ization of gpu tee overheads in distributed data parallel ml training, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":22,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:36.423186Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:d01b7689f155e68bd7007baa5071c839fa104fdcb62a9032adb20550fd248e43","observation_id":"fa13cddd-7e45-47ef-8c36-08f2571bb453","resolution":{"observed_at":"2026-08-02T18:40:36.423186Z","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":"2026-08-02T18:40:36.524790Z","title":"Verilora: Fine-tuning large language models with verifiable security via zero-knowledge proofs, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":23,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:36.524790Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:dbc4a01fe6b57d37324f221182e0e636e260903a86f1ab3e1fe866a36a76ef5b","observation_id":"aa6c5f58-acce-40a5-8d37-4b9348a20c29","resolution":{"observed_at":"2026-08-02T18:40:36.524790Z","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":"2026-08-02T18:40:36.642521Z","title":"Against the achilles’ heel: A survey on red teaming for generative models.Journal of Artificial Intelligence Research 82(2025), 687–775","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":24,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:36.642521Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:fa05cf91bed968cefd20f03084212b55e21b989a562105a7ec329d3be1d35366","observation_id":"c89a2de4-294e-4813-a862-1bc399308c2a","resolution":{"observed_at":"2026-08-02T18:40:36.642521Z","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":"2026-08-02T18:40:36.743782Z","title":"Towards deep learning models resistant to adversarial attacks, 2019","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":25,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:36.743782Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:45588fdcef3a21dd0081f7910708692b21c62552e6e3a007d7a94fed0b675f96","observation_id":"249402ea-0e02-4838-ac5b-96aea6bf2082","resolution":{"observed_at":"2026-08-02T18:40:36.743782Z","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":"2026-08-02T18:40:36.841734Z","title":"Ppfl: Privacy-preserving federated learning with trusted execution environments","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":26,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:36.841734Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:4a2b5fe848637d25987065537495288e6c6b93bb3aaca79aa9cb727c788ed04d","observation_id":"4a1b2e27-9c7a-4ed4-9b6d-434ce26d9628","resolution":{"observed_at":"2026-08-02T18:40:36.841734Z","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":"2026-08-02T18:40:37.000004Z","title":"Introducing improvements to the fine-tuning api and expand- ing our custom models program","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":27,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:37.000004Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:14e8bb1ce67d2cedfc25306ba096a9ff9a1c2a0531e9ce4c2ecd20a1a0ddcf15","observation_id":"2df9c398-2e39-4fe1-af63-f4757babcec7","resolution":{"observed_at":"2026-08-02T18:40:37.000004Z","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":"2026-08-02T18:40:37.125989Z","title":"Model optimization — openai api guides, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":28,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:37.125989Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:be65c0e3f475c554b888faca5ac39dc6b0486e407b1c216cb01cdda5c8b6e38c","observation_id":"33ce46c9-bb8c-4748-b4a4-6108881ca0d0","resolution":{"observed_at":"2026-08-02T18:40:37.125989Z","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":"2026-08-02T18:40:37.231993Z","title":"Dinov2: Learning robust visual features without supervision, 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":29,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:37.231993Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:93152d5480716e8dbb9026cc7559f01875c7cb178b128a526eba778a27321e39","observation_id":"fd6a339e-9b0e-4799-a569-28a71f9118e4","resolution":{"observed_at":"2026-08-02T18:40:37.231993Z","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":"2026-08-02T18:40:37.295425Z","title":"zkgpt: An efficient non-interactive zero-knowledge proof framework for llm inference","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":30,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:37.295425Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:b9aa5bf7bf49402abecd11079f2a6dcd3bd6454454e9e10bf8493f7f1ae86d70","observation_id":"175aecb9-5011-4830-97d0-15213d81a17c","resolution":{"observed_at":"2026-08-02T18:40:37.295425Z","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":"2026-08-02T18:40:37.387259Z","title":"L., GREGOR, F., ARNAUTOV, S., KUNKEL, R., BHATOTIA, P.,ANDFETZER, C","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":31,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:37.387259Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:18b08e31710f03f86f49a0284b45aae255ee36ae186182c0f20e5a68fa0a6452","observation_id":"0b9a59d3-0f18-4ac0-9879-2e9e70e7367d","resolution":{"observed_at":"2026-08-02T18:40:37.387259Z","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":"2026-08-02T18:40:37.539982Z","title":"Zero- knowledge ai inference with high precision.ACM CCS(2025)","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":32,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:37.539982Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:6c06e498156ffeba1567313b474ac69e940b2d9af10f41ad9b4da3d78b0a7648","observation_id":"d19cee77-899c-4d49-b057-1c47173cb6b1","resolution":{"observed_at":"2026-08-02T18:40:37.539982Z","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":"2026-08-02T18:40:37.631452Z","title":"C.,ANDFEI-FEI, L","venue":null,"work_id":null,"year":2015},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":33,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:37.631452Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:257d10b958ad6e630a319ccb3d7f44646c8d1154966d608208c01f165dab4d52","observation_id":"ed8c5432-2c1e-42e6-a0c2-2c53e21b01a4","resolution":{"observed_at":"2026-08-02T18:40:37.631452Z","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":"2026-08-02T18:40:37.741620Z","title":null,"venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":34,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:37.741620Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:f598c1fd74f008045463890f4bd76de56cf08a6243283326ba44dc3608c668a8","observation_id":"4e8950a5-5523-4f95-8aa0-f745372e3b3e","resolution":{"observed_at":"2026-08-02T18:40:37.741620Z","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":"2026-08-02T18:40:37.923780Z","title":"Compact proofs of retrievability.J","venue":null,"work_id":null,"year":2013},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":35,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:37.923780Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:4d0a0f21f6b419904ee300df11a1bdd857e5f98f508083c9a9a6d2ff9ac8351f","observation_id":"f0d3c508-47ef-4064-b6a4-6610b51be15b","resolution":{"observed_at":"2026-08-02T18:40:37.923780Z","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":"2026-08-02T18:40:38.067919Z","title":"C., BIRKE, R.,ANDPERRI, S","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":36,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:38.067919Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:849e19aa153557656cb0b057682c3f11fb8c82417dc8af024def6f7a5230014f","observation_id":"f615bee5-6d2f-4d78-b69a-c821bb33562d","resolution":{"observed_at":"2026-08-02T18:40:38.067919Z","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":"2026-08-02T18:40:38.215985Z","title":"Trusted yet flexible: High- level runtimes for secure ml inference in tees.Journal of Cybersecurity and Privacy 6, 1 (2026)","venue":null,"work_id":null,"year":2026},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":37,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:38.215985Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:dac59a377a6b896c95fb73bb286d98c3d52340f5a76bc55093f26f193029f5a3","observation_id":"e9196c40-33b3-425f-a745-ba666a7e7f70","resolution":{"observed_at":"2026-08-02T18:40:38.215985Z","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":"2026-08-02T18:40:38.370818Z","title":"zkllm: Zero knowledge proofs for large language models","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":38,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:38.370818Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:b1e3bd3bd01d9672faedebbbffa7e262dde56e3bc9dab9ab9d0909124226215c","observation_id":"a6194aca-fddf-455e-8fd4-834b90ebf5ad","resolution":{"observed_at":"2026-08-02T18:40:38.370818Z","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":"2026-08-02T18:40:38.523053Z","title":"Svip: Towards verifiable inference of open-source large language models.arXiv preprint arXiv:2410.22307(2025)","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":39,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:38.523053Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:4e641fe912af9eb5a3e321ee1e8314fe5d1e84129c720aedcf0ea5385d1a6e15","observation_id":"17cbbcfb-a0bf-477d-a699-26ef0efbf2c6","resolution":{"observed_at":"2026-08-02T18:40:38.523053Z","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":"2026-08-02T18:40:38.710020Z","title":"Pipellm: Fast and confiden- tial large language model services with speculative pipelined encryption","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":40,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:38.710020Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:b8ae4c29403ef4689ffed0fbc7cd72991f9a9d7aa15df096b5712a5a2a4e237f","observation_id":"044b0072-0bd3-4c67-a37f-98e238d7a9d4","resolution":{"observed_at":"2026-08-02T18:40:38.710020Z","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":"2026-08-02T18:40:38.857206Z","title":"H2:towards efficient large-scale llm training on hyper-heterogeneous cluster over 1,000 chips, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":41,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:38.857206Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:4e2d2f04256054c269ce98c94e39e178663296df4029f53371cef37731158020","observation_id":"0f27bae9-8dc6-4d4b-a661-84c9ea38ac37","resolution":{"observed_at":"2026-08-02T18:40:38.857206Z","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":"2026-08-02T18:40:39.003180Z","title":"Visual transform- ers: Token-based image representation and processing for computer vision, 2020","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":42,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:39.003180Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:ccfc52309d0b6ea8228ed9eb4fc639e46a8457d27ead8b8ecd45e50d2e53366e","observation_id":"a2edff27-3c07-4460-822b-935ed5893411","resolution":{"observed_at":"2026-08-02T18:40:39.003180Z","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":"2026-08-02T18:40:39.050167Z","title":"zkpytorch: A hierarchical optimized compiler for zero-knowledge machine learning.Cryptology ePrint Archive(2025)","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":43,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:39.050167Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:045d2f56af57e46ed1956ba554d626001822ce8f6baf63f1d4693d0a61db5cd4","observation_id":"d6236040-7e9e-4b1c-a780-28a0457d5317","resolution":{"observed_at":"2026-08-02T18:40:39.050167Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2503.13509","last_updated":"2025-06-02T02:53:02Z","snapshot_observed_at":"2026-08-07T17:05:43.567554Z","submitted_at":"2025-03-13T20:25:10Z","title":"MentalChat16K: A Benchmark Dataset for Conversational Mental Health Assistance","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2503.13509","snapshot_observed_at":"2026-08-02T18:40:39.176313Z","title":"Men- talchat16k: A benchmark dataset for conversational mental health as- sistance.arXiv preprint arXiv:2503.13509(2025)","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":44,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:39.176313Z"},"links":{"cited_paper":"/paper/2503.13509","citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:4704d3978f900f2c628fd96e04738bc07a83ee440caccbd86ec8d33785a10cba","observation_id":"419f79a0-6dba-43e1-b234-67624cd332b8","resolution":{"observed_at":"2026-08-02T18:40:39.176313Z","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":"2026-08-02T18:40:39.251168Z","title":"Hex- iscale: Accommodating large language model training over heteroge- neous environment, 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":45,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:39.251168Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:815d70c2e4bab718de72face6b91bb30c01865e38692fe3f4ed2ea6700459391","observation_id":"bc2d74de-41c5-4d82-a4d4-56595c8d63c4","resolution":{"observed_at":"2026-08-02T18:40:39.251168Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2412.15115","last_updated":"2025-01-03T02:18:21Z","snapshot_observed_at":"2026-07-06T02:11:23.670680Z","submitted_at":"2024-12-19T17:56:09Z","title":"Qwen2.5 Technical Report","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2412.15115","snapshot_observed_at":"2026-08-02T18:40:39.349252Z","title":"Qwen2.5 technical report.arXiv preprint, arXiv:2412.15115 (2025)","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":46,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:39.349252Z"},"links":{"cited_paper":"/paper/2412.15115","citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:42104e1b7cf47a49749a08067cb9c8b93ad1daf39ccadde40aaa713fa437a0f0","observation_id":"4d775fcd-22fd-433b-9606-3affa157c44d","resolution":{"observed_at":"2026-08-02T18:40:39.349252Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2411.06611","last_updated":"2024-11-12T03:04:07Z","snapshot_observed_at":"2026-07-06T19:48:10.800324Z","submitted_at":"2024-11-10T22:08:37Z","title":"vTune: Verifiable Fine-Tuning for LLMs Through Backdooring","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2411.06611","snapshot_observed_at":"2026-08-02T18:40:39.455371Z","title":"vtune: Ver- ifiable fine-tuning for llms through backdooring.arXiv preprint arXiv:2411.06611(2024)","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":47,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:39.455371Z"},"links":{"cited_paper":"/paper/2411.06611","citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:204d6101afd3412594bdd8efde3f0773c60bc98c02e37230a2e90e303b37486f","observation_id":"3799bfb7-6132-4847-8641-de8254501949","resolution":{"observed_at":"2026-08-02T18:40:39.455371Z","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":"2026-08-02T18:40:39.550384Z","title":"En- abling execution assurance of federated learning at untrusted partic- ipants","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":48,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:39.550384Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:63ed102a3efc5b5fbde721af565a24086be556a39ed7e06ba6ef1d0bd48d6cae","observation_id":"ff8b25fe-c989-4072-b458-22d5afdfd9b8","resolution":{"observed_at":"2026-08-02T18:40:39.550384Z","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":"2026-08-02T18:40:39.638638Z","title":"Sear: Secure and efficient aggregation for byzantine-robust federated learning.IEEE Transactions on Dependable and Secure Computing 19, 5 (2021), 3329–3342","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI","version":2},"reference_index":49,"source":"pdf_text","source_observed_at":"2026-08-02T18:40:39.638638Z"},"links":{"citing_paper":"/paper/2603.07466"},"observation_digest":"sha256:fd2de1a247d174ff441fada322c9ba964eb5c44fade729e5a4911d90d8e39259","observation_id":"e5200c61-c990-4c74-8283-4e27978c342c","resolution":{"observed_at":"2026-08-02T18:40:39.638638Z","resolver_source":null,"status":"malformed_identifier"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}}],"paper":{"arxiv_id":"2603.07466","last_updated":"2026-07-22T07:08:12Z","latest_version":2,"primary_category":"cs.CR","snapshot_observed_at":"2026-08-02T18:40:32.079175Z","submitted_at":"2026-03-08T05:06:45Z","title":"Trusting What You Cannot See: Auditable Fine-Tuning and Inference for Proprietary AI"},"reference_resolution":{"displayed":49,"state_counts":{"malformed_identifier":1,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":48,"verified_exact":0,"verified_fuzzy":0},"total_outbound_references":49},"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":"2026-08-08T06:32:00.761636+00:00","source":"crossref"},{"observed_at":"2026-08-08T06:31:55.24221+00:00","source":"retraction_watch"}],"thesis":"As of 8 August 2026, this Paper Citation Record lists 49 of 49 outbound references and 2 inbound Pith citation observations for arXiv:2603.07466."}