AI Disclosure

5.5 AI contributors and reviewers

Compiled by Muse Spark 1.3. Each model is named as credited: where the original gives no release number, the table says so rather than inferring one. "Original" roles are in the 2025 publications; audit, correction, review, and checking roles are in the collection's preparation. Conflict and recusal statements for each record are in Appendix E. The table's sources, version history, and identification notes are in Appendix G.

Model (as credited) Contribution Records
Muse Spark 1.3 Author of 2 new additions; reviewer of 4 new additions and 1 corrected text; first auditor of 3 parents; second auditor of 10 parents and independent verifier of 1; corrections preparer for 5 parents; contributor to the editorial standard and its consolidation; registrar of the Master Register; assembly handoff for an early draft of the volume; transcriber and rebuilder of the 4 full-post transcriptions; adjudicator of the second check of the four; issuer of the rebuilt Mixture of Experts archive text and the Lexicon (4) archive text; verifier of a draft of the volume; checker of the Appendix F archive text and of the Lexicon (6) archive text at anchor level; compiler of this table, the consolidated bibliography, and the title and copyright pages; co-author of the Volume Preface; comparer of the blind audit of Circuit Sparsity; auditor of drafts of the volume and reconciler of GPT-6's final whole-volume review; reviewer and checker of the Annual Introduction's first section; preparer of the published-formula restoration texts and the archive check statements; preparer of the bibliography verification report and of the reader-text sweep edit list for the four recused texts; co-designer of the cover and logo (concept by the editor) Author: AIK-2025-F-N-RLVR, AIK-2025-F-N-CONTEXTENG. Reviewer: AIK-2025-F-N-CONTEXTROT, AIK-2025-F-N-EMERGENTMIS, AIK-2025-F-N-COTMONITOR, AIK-2025-F-N-SCHEMING, AIK-2025-F-P-COTPROMPT. First audit: AIK-2025-F-P-MOE, AIK-2025-F-P-CONSTAI, AIK-2025-L-P-LEX2025N3 (+ 12 records). Second audit: AIK-2025-F-P-CALM, AIK-2025-F-P-PDDISAGG, AIK-2025-F-P-GENUI, AIK-2025-F-P-ONPOLDISTIL, AIK-2025-L-P-LEX2025N2 (+ 12), AIK-2025-F-P-MOD, AIK-2025-L-P-LEX2025LAUNCH (+ 26), AIK-2025-L-P-LEX2025N5 (+ 12), AIK-2025-L-P-LEX2025N7 (+ 12), AIK-2025-L-P-LEX2025N4 (+ 12); independent verification: AIK-2025-F-P-COTPROMPT. Corrections: AIK-2025-F-P-MOE, AIK-2025-F-P-CONSTAI, AIK-2025-L-P-LEX2025N3, AIK-2025-L-P-LEX2025N4, AIK-2025-F-P-CIRCUITSP. Transcriptions: posts 13958, 13998, 14221, 14343 (rebuilt from stored content). Whole volume: editorial standard, register, assembly, bibliography, this table; Volume Preface (co-author); blind-audit comparison (Circuit Sparsity); draft audits and final-review reconciliation; Annual Introduction first-section review and check; formula restoration and archive check statements; bibliography verification report; reader-text sweep edit list for the four recused texts; cover and logo (co-designer)
Claude Opus 5.5 Author of 1 new addition; reviewer of 4 new additions and 1 installment's corrected records; reviewer of corrections /C07 and /C08 to Lexicon (4); first auditor of 13 posts: 12 canonical parents and excluded Lexicon (6); corrections preparer for 10 parents; volume assembly; author of the Annual Introduction's second section ('The Year in AI Terms, 2025') and drafter of the Editorial Method, and of the Appendix E summary; reviewer of the Annual Introduction's first section (reports filed); drafter of the title and copyright pages; drafter of the Appendix D digest addendum; drafter of the review briefs; co-author (drafter) of the Volume Preface; author of the checking briefs; transcriber of 11 of the 12 archive texts (all except Lexicon (4)) and of the Appendix F archive text; corrected the Chain-of-Thought Prompting and Lexicon (2) archive texts; author of the identifier mapping; author of the reader-text sweep proposal and line list and of the bibliography reconciliation report. Recused from every text authored by an Anthropic model, except the first audit of Circuit Sparsity, the exception recorded in Appendix E Author: AIK-2025-F-N-CONTEXTROT. Reviewer: AIK-2025-F-N-RLVR, AIK-2025-F-N-EMERGENTMIS, AIK-2025-F-N-COTMONITOR, AIK-2025-F-N-SCHEMING, AIK-2025-L-P-LEX2025N4 (+ 12, including corrections /C07 and /C08). First audit: AIK-2025-F-P-COTPROMPT, AIK-2025-F-P-CALM, AIK-2025-F-P-PDDISAGG, AIK-2025-F-P-GENUI, AIK-2025-F-P-ONPOLDISTIL, AIK-2025-L-P-LEX2025N2 (+ 12), AIK-2025-F-P-MOD, AIK-2025-L-P-LEX2025LAUNCH (+ 26), AIK-2025-L-P-LEX2025N5 (+ 12), AIK-2025-L-P-LEX2025N7 (+ 12), AIK-2025-L-P-LEX2025N4 (+ 12), AIK-2025-F-P-CIRCUITSP, post 14313 (New AI Lexicon 2025 (6), excluded, no Public Entry ID). Corrections: 10 of the 12 canonical parents above, except AIK-2025-L-P-LEX2025N4 and AIK-2025-F-P-CIRCUITSP (both Muse). Transcriptions: 11 of the 12 archive texts (all except post 14233, Lexicon (4)) and post 14267 (Appendix F); corrections to the COTPROMPT and L2 archive texts. Whole volume: assembly, mapping, checking briefs, review briefs, reader-text sweep proposal and line list, bibliography reconciliation report, front matter (Annual Introduction, second section; Editorial Method; Appendix E summary; title and copyright pages), Appendix D digest addendum; Volume Preface (co-author, drafter); Annual Introduction first-section reviews (reports filed)
Claude Fable 5.2 Checker of a policy draft; link-checker of the consolidated bibliography; recorder of the editor's rulings on the Volume Preface and two identifications; preparer of the front-matter-and-checks package; volume assembly. Recused from every text authored by an Anthropic model (Appendix E), as is Claude Opus 5.5 Whole volume: provenance-policy check; bibliography link check; rulings record; package preparation; volume assembly
GPT-6 (OpenAI), release GPT-6 Astra Author of the editorial standard; author of 3 new additions; reviewer of 3 new additions, 1 corrected text, and 1 installment's corrected records; reviewer of drafts of the volume (including the final whole-volume review; report filed); changes-only checker of the editorial-standard candidate; reviewer of the provenance policy AP-2025-01; co-author of the Volume Preface; co-designer of the cover and logo (concept by the editor) Author: AIK-2025-F-N-EMERGENTMIS, AIK-2025-F-N-COTMONITOR, AIK-2025-F-N-SCHEMING. Reviewer: AIK-2025-F-N-RLVR, AIK-2025-F-N-CONTEXTENG, AIK-2025-F-N-CONTEXTROT, AIK-2025-F-P-GENUI, AIK-2025-L-P-LEX2025N2 (+ 12). Whole volume: editorial standard, draft reviews (including the final whole-volume review), standard-candidate check; provenance-policy review (AP-2025-01); Volume Preface (co-author); cover and logo (co-designer)
GPT-6.2 Astra (OpenAI) Author of the Annual Introduction's first section ("What AI Learned to Do in 2025 and How We Could Tell"); reviewer of a proof of the volume (report filed); author of the publication audit of a later proof (report filed) Whole volume: Annual Introduction, first section; proof review (report filed); publication audit of a later proof (report filed)
Claude Sonnet 4.5 Original author of 2 standalone entries and 1 installment's entries, and of 1 entry's base text state; co-author of 1 entry's revised drafts and final version; commenter on 1 entry; original reviewer of 1 entry and named reviewer of 2 entries (no Claude text in either post) Author: AIK-2025-F-P-CONSTAI, AIK-2025-L-P-LEX2025N3 (+ 12), AIK-2025-F-P-CIRCUITSP (T2/V2, the base version). Co-author: AIK-2025-F-P-MOE (revised drafts and final version, with ChatGPT). Comments: AIK-2025-F-P-ONPOLDISTIL. Reviewer: AIK-2025-F-P-MOD (of v2, plus a post-publication note on v3); named reviewer: AIK-2025-F-P-CALM, AIK-2025-F-P-PDDISAGG
ChatGPT (release not stated) Original author, co-author, and reviewer across 8 parents and 1 excluded installment, as credited without a release number. The mapping identifies some ChatGPT texts as GPT-5; those are listed separately below Co-author: AIK-2025-F-P-MOE (revised drafts and final version, with Claude Sonnet 4.5). Author: AIK-2025-F-P-PDDISAGG (final version). Embedded reviews: AIK-2025-L-P-LEX2025N3 (two reviews, not applied), AIK-2025-L-P-LEX2025N5, AIK-2025-L-P-LEX2025N7, AIK-2025-L-P-LEX2025N4. Reviewer: AIK-2025-F-P-COTPROMPT (review wording carried into v3), AIK-2025-F-P-CONSTAI, AIK-2025-F-P-MOD (of v1). Archive only: post 14313, Lexicon (6), with Qwen3-Max (excluded, disposition X)
ChatGPT (GPT-5) Original author, identified as GPT-5 by the mapping or credited in the post as "ChatGPT-5" Author: AIK-2025-F-P-ONPOLDISTIL (entry), AIK-2025-L-P-LEX2025LAUNCH (+ 26), AIK-2025-F-P-MOD (v2 and v3, the text of record)
Grok (release not stated) Original reviewer of 2 entries; fact-checker of 1 entry, as credited without a release number Reviewer: AIK-2025-F-P-COTPROMPT, AIK-2025-F-P-CONSTAI. Fact-check: AIK-2025-F-P-ONPOLDISTIL
Grok (Expert) Original reviewer, as stated in the post Reviewer: AIK-2025-F-P-MOE
Grok 4 Expert Original reviewer of a draft and author of a separate text state in 1 entry (the review was not applied to the base version) AIK-2025-F-P-CIRCUITSP (review; T4/V4)
Grok 4.1 Original author of 1 installment's texts (byline "Grok" with no release number; the exchange addresses Grok 4.1) AIK-2025-L-P-LEX2025N4 (+ 12)
Grok 4.5 Reviewer of 2 installments' corrected records; reviewer of the correction to LExI (Launch Edition; report filed); reviewer of the correction to Test-Time Compute Scaling (Lexicon (3); report filed); archive checker of 2 archive texts (accepted) and of the 4 full-post transcriptions (checked; findings largely not sustained on adjudication) Reviewer: AIK-2025-L-P-LEX2025N5 (+ 12), AIK-2025-L-P-LEX2025N7 (+ 12), AIK-2025-L-P-LEX2025LAUNCH (correction to LExI; report filed), AIK-2025-E-P-TTCSCALE (correction /C07; report filed). Checking: AIK-2025-F-P-MOD, AIK-2025-F-P-GENUI (accepted); posts 13958, 13998, 14221, 14343 (report filed; adjudicated — two findings survive independently)
Gemini 2.5 Pro Original writer of 1 entry; original reviewer of 1 entry Writer: AIK-2025-F-P-COTPROMPT. Reviewer: AIK-2025-F-P-MOE
Gemini 3 Pro Sole contributor to 1 entry; contributor to 1 installment AIK-2025-F-P-GENUI (sole contributor); AIK-2025-L-P-LEX2025N2 (+ 12)
Gemini Thinking with 3 Pro Author of 1 text state in 1 entry AIK-2025-F-P-CIRCUITSP (T3/V3)
Gemini 3 Pro Thinking Contributor to the Appendix F post (related editorial material) Post 14267 (Appendix F), with GPT-5 via Copilot Smart
DeepSeek-V3.2 Original author of 2 entries' drafts and 1 installment's texts (credited "DeepSeek-3.2" in the installment); original reviewer of 2 entries Author: AIK-2025-F-P-CALM (initial draft), AIK-2025-F-P-MOD (v1), AIK-2025-L-P-LEX2025N5 (+ 12). Reviewer: AIK-2025-F-P-MOE, AIK-2025-F-P-MOD (accepted v2)
DeepSeek-V4.1 Reviewer of 4 corrected texts and 2 installments' corrected records; assigned archive checker — returned no checks (role reassigned to GLM 5.3 Flash). Reviewer of the volume Introduction and the Editorial Method (reports filed) Reviewer: AIK-2025-F-P-COTPROMPT, AIK-2025-F-P-CONSTAI, AIK-2025-L-P-LEX2025N3 (+ 12), AIK-2025-F-P-ONPOLDISTIL, AIK-2025-L-P-LEX2025LAUNCH (+ 26), AIK-2025-F-P-CIRCUITSP. Assigned checking, not returned: posts 13958, 13998, 14221, 14343, and 8 archive sections (package-consistency note filed, no live-post check). Whole volume: Introduction and Method reviews (reports filed)
GLM 5.3 Flash Reviewer of 4 corrected texts (second check and independent review); archive checker of 7 archive sections and 1 cross-sample; second checker of the 4 full-post transcriptions (report filed; findings led to the rebuild); blind first auditor of 1 entry (report filed; 13 findings, all recorded as annotations, corrections stand); checker of the rebuilt Mixture of Experts transcription (accepted, no discrepancies); comparer of the 4 PDF archive sections against the online transcriptions (report filed); checker of the Annual Introduction's first section (report filed) AIK-2025-F-P-MOE, AIK-2025-F-P-CALM, AIK-2025-F-P-PDDISAGG, AIK-2025-F-P-MOD. Checking: AIK-2025-F-P-ONPOLDISTIL, AIK-2025-F-P-COTPROMPT, AIK-2025-F-P-PDDISAGG, AIK-2025-L-P-LEX2025LAUNCH (+ 26), AIK-2025-L-P-LEX2025N4 (+ 12), AIK-2025-L-P-LEX2025N2 (+ 12), AIK-2025-L-P-LEX2025N7 (+ 12) — plus the AIK-2025-F-P-GENUI cross-sample. Four transcriptions (report filed). Blind first audit: AIK-2025-F-P-CIRCUITSP (report filed; comparison by Muse Spark 1.3). Rebuilt Mixture of Experts transcription (post 13998): accepted, no discrepancies. Archive comparison: posts 13958, 13998, 14221, 14343 (report filed). Annual Introduction, first section: checked (report filed)
Qwen3-Max Author of 1 entry's initial draft; co-author of 1 excluded installment (archive only); archive checker of 2 archive texts (accepted, reports filed); checker of 3 rebuilt transcriptions (wording accepted, with points marked not verified in Lexicon (3) and Circuit Sparsity) and of the last third of the COTPROMPT archive text (accepted) AIK-2025-F-P-MOE (initial draft). Archive only: post 14313, Lexicon (6), with ChatGPT (excluded, disposition X). Checking: AIK-2025-F-P-CALM, AIK-2025-L-P-LEX2025N5 (+ 12) (accepted, reports filed). Checked: rebuilt transcriptions of posts 13958, 14221, 14343 — wording accepted, points marked not verified in Lexicon (3) and Circuit Sparsity; COTPROMPT A-239–A-293 (accepted)
Qwen4-Max (release number not exposed) Reviewer of the volume Introduction and the Editorial Method (reports filed); reviewer of a proof of the volume (whole-volume review; report filed); assigned blind first auditor of 1 entry — stopped before auditing under the brief's contamination protocol, no coverage, no findings Whole volume: Introduction reviews, Editorial Method review, proof review (report filed). Assigned audit, stopped with no coverage: AIK-2025-F-P-CIRCUITSP
Kimi (Moonshot AI; release not stated) Adjudicator of two challenged audit findings on the published formulas in Mixture of Experts and New AI Lexicon 2025 (3) (report filed): both not upheld — the formulas are present as formula images, and the archive gaps are text-only capture losses AIK-2025-F-P-MOE; AIK-2025-E-P-TTCSCALE (via parent AIK-2025-L-P-LEX2025N3)
Kimi K2 Author of 1 installment's list texts AIK-2025-L-P-LEX2025N7 (list texts)
Kimi K2 Thinking Author of 1 entry's initial version (off-post) AIK-2025-F-P-PDDISAGG (initial version)
GPT-5 Mini Author of 1 entry's final version AIK-2025-F-P-CALM (final version)
GPT-5.2 Author of 1 text state in 1 entry AIK-2025-F-P-CIRCUITSP (T1/V1)
Copilot Smart, GPT-5.2 Author of 1 installment's full texts and a review, as credited jointly in the register AIK-2025-L-P-LEX2025N7 (full texts and a review)
GPT-5 via Copilot Smart Contributor to the Appendix F post (related editorial material) Post 14267 (Appendix F), with Gemini 3 Pro Thinking

Counts. 29 credited model identities: 18 in the canonical text's original publications and front matter, 2 in the appendix only (Gemini 3 Pro Thinking, GPT-5 via Copilot Smart), and 9 whose roles are entirely in the collection's preparation (Muse Spark 1.3, Claude Opus 5.5, Claude Fable 5.2, GPT-6, DeepSeek-V4.1, Grok 4.5, GLM 5.3 Flash, Qwen4-Max, Kimi). Every parent record's contributor and reviewer lines in the Master Register are represented above; contained records inherit their parent's line. No release number has been inferred: "release not stated" means the post, the mapping, and the register give none. The 107 register rows are 101 records plus 6 installment containers.

5.6 AI disclosure and conflicts of interest

AI disclosure. AI models wrote this volume under the direction of a human editor. The 2025 posts were written by models from several providers, and in most posts reviewed by models from other providers, in dialogues the editor facilitated. Models also carried out the collection's preparation: inventory, audits, corrections, reviews, archive checks, registration and assembly. The Annual Introduction and this section were drafted by Claude Opus 5.5. DeepSeek-V4.1 and Qwen4-Max reviewed the Introduction at v0.1 and v0.2.1 and this section at v0.1, and both reviewed this section at v0.2.3 and the Introduction's changes at v0.3.2 (November 16, 2026); DeepSeek-V4.1's reviews reached the builder through the editor and were accepted by the editor's direction of November 16, 2026, and the registrar does not hold them; the other front matter names its own authors. The editor, W.H.L., took the decisions: every disposition, every approval, every ruling. A reviewer's verdict is a recommendation to the editor, never approval.

Providers. The models that prepared the collection, with their providers: Claude Opus 5.5 and Claude Fable 5.2 (Anthropic); Muse Spark 1.3 (Meta); GPT-6 (OpenAI); DeepSeek-V4.1 (DeepSeek); Qwen4-Max and Qwen3-Max (Alibaba's Qwen team); Grok 4.5 (xAI); GLM 5.3 Flash (Z.ai; served by Mistral AI as the editor accessed it). The providers of the models in the 2025 posts are given in §5.5.

Independence. The rule is that no model audits (first or second), reviews or checks a text in which a model from its own provider took part: wrote, co-wrote or reviewed it, in the 2025 post or in its correction. (That the rule covers first audits is written into the provenance policy, AP-2025-01 §9, effective on its approval; the Circuit Sparsity exception below is why.) A model whose review was printed in a post counts as having taken part, whether or not the review was applied. Lexicon installments are reviewed as one package, so the rule applies to every provider in the installment. The rule governs the 2025 texts, their corrections and their archive transcriptions. The commissioned entries were reviewed by models from providers other than the drafter's (§5.3). The front matter's reviewers were appointed by the editor, with their providers' roles disclosed in their reports. Two exceptions are on the record:

Recusal. Claude Opus 5.5, which assembled the volume, and Claude Fable 5.2, which held the same role for part of November 16, 2026, are Anthropic models. Both are recused from the texts written in whole or in part by Anthropic models: the four named above. For those texts they reviewed procedure only (identifiers, hashes, format) and took no view on their content or corrections, apart from the Circuit Sparsity first audit described above.

Disclosure. Each model that reviewed or checked a text was asked to disclose in its report any other contact with that text, and the disclosures are recorded as reported. The conflict statements in Appendix E are drawn from the Master Register, not from self-report alone. §5.5 lists each model's roles; Appendix E gives the conflict and recusal statement for each record, and a summary of the practice.