Sources we considered, with review status and reasons.
3070 recorded entries · 652 awaiting detailed review · 584 excluded · 672 on hold · 3 publisher withdrawals · 1159 related to curated records.
Suggest a missing contribution or correction.
See omissions and decisions reviewed against public discussion.
Entries without a recorded subject appear only under All subjects. Coverage totals describe all subjects.
1911 matching entries · Page 14 of 39
The original extension-conjecture paper reports a mathematical result but the available HTML has no AI-origin disclosure, so the model’s role cannot be determined.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The Erdős #593 original problem record returned 403, so the claimed solution and its AI origin could not be checked.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The community record says GPT-5.6 Sol chose the problem and produced a short two-lemma proof, but the sole cited original X post was not available here and the canonical conjecture wording has two readings.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The dimension-two paper derives its map from an earlier ChatGPT-associated three-variable construction, but its provenance does not identify who found the plane reduction or new proof.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The original Erdős problem page returned 403, leaving no readable primary proof or AI disclosure for this entry.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The registry describes an AI-found finite counterexample, but the original X announcement and shared chat could not be read here; central AI origin and exact certificate remain unconfirmed.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The cited formal-conjectures pull request could not be read; a formalized conjecture alone would not establish either a proof or central AI-origin mathematical work.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The disclosure lists ChatGPT/Codex exploration and verification code among several roles, without identifying a central AI-origin counterexample or proof idea.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Fitzsimons attributes Codex and Aristotle to Lean formalization and proof search and Codex to editorial suggestions. The paper does not credit either with the transcript-envelope mechanism or the pure-DP query-release theorem.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The community entry credits a Grok agent with a Hoffman–Singleton counterexample, but its discussion alleges an earlier GPT-assisted solution with a separate Lean repository; the original X certificate could not be read here.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The source does not distinguish a retained central ChatGPT idea from assistance with the author’s code-inspired construction.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The announcement gives an exact graph and thanks two models and human reviewers, but does not say who found the counterexample or designed the search.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The manuscript says GPT-5.6 helped develop earlier operator-algebra arguments, but the author rewrote and streamlined that section. It does not identify which, if any, central AI idea survives in the final spectral-edge proof.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The disclosure expressly credits the core proof ideas and overall strategy to the human authors. ChatGPT generated search code and assisted computations, symbolic derivations, and prose.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The paper says Codex helped with proof exploration and editorial assistance, without attributing a central polynomiality argument to it.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The paper acknowledges ChatGPT collaboration but does not identify a retained central model-origin step in the categorical cocycle result.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The source reports GPT use for proof preparation but does not identify which central two-leaf argument originated with the model.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-19T00:00:00.000Z
The authors report ChatGPT assistance with the counterexample but do not identify which central construction or calculation originated with the model.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The author’s AI-use record says Zachary J. Lewis developed the central ideas, proof strategy, and arguments; GPT aided exploration, checks, exposition, and Lean work.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Metadata screened. Requires original-source review of the specific contribution, baseline and central AI work, with separate AI involvement, technical verification, human understanding and importance ratings before publication.
Screening recorded: 2026-10-27
The current disclosure says model input was indispensable but does not identify a retained central model-supplied construction or sharp-constant argument.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The Erdős proof-claim forum returned 403. The Lean file establishes an artifact but does not by itself identify who found the mathematical argument or whether AI originated it.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Gao’s title-page disclosure says the main idea and framework are his and Claude Fable 5 assisted proofs and writing. The abstract also credits Alpöge’s prior dimension-three counterexample; no central AI-origin contribution is identified for this paper.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The original hypergraph Moore-bound paper establishes the theorem, but a primary disclosure of central AI work was not located in the accessible manuscript.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The acknowledgments report Codex and Claude Code for proof exploration and criticism, but do not assign the linear L3 bound or a retained decisive argument to either model.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Sawin credits ChatGPT with a pattern observation and change of variables used in the proof, while he describes the restriction and adapted counting strategy as his own. Whether the model’s change of variables is central enough for inclusion needs a closer proof comparison.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The authors credit ChatGPT with searching and refining the construction but do not identify a retained model-origin choice after the human-supplied starting idea.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-19T00:00:00.000Z
The paper says Codex and Claude Code assisted proof exploration and criticism for long rainbow arithmetic progression bounds, but gives no model-origin lemma or construction.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The paper reports broad model help in exploration and proof development but does not identify a central retained AI idea for the final proof.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The tomography authors thank ChatGPT for discussing and refining proof ideas and presentation, while assigning responsibility for results to themselves; the central lower-bound construction is not attributed to AI.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The Kac-walk paper reports a cutoff theorem, but the accessible original text does not identify central AI-origin work.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The source says ChatGPT worked under the author’s direction but does not identify which formula or derivation originated with the model.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The disclosure says AI-assisted tools developed search and verification programs and a Lean formalization under author direction, but does not say whether AI originated the central Hamiltonicity argument.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The directory lists an AI-related resolution of Erdős #336, but the cited problem record could not be accessed, leaving the original proof and model-origin role unconfirmed.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The disclosure credits ChatGPT with literature search and accelerating a search, but not with the retained central operator-system construction.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-19T00:00:00.000Z
The queue points to Erdős problem 538, but its linked problem record and a specific AI-origin proof were not available in the reviewed source evidence.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The directory links Erdős #267 to an AI-related solution, but its sole original problem record could not be accessed and no attributable central model step could be established.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The original problem page describes covering-system estimates but supplies no AI-origin proof or contribution for this entry.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The primary note argues the printed Erdős 662 wording is inconsistent and reconstructs a threshold statement related to an already known two-smallest-distances theorem. Its model provenance and exact new result are not documented in the note.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The original Erdős record could not be read, so the precise result and its AI disclosure remain unknown.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The directory lists an AI-related result for Erdős #254, but the sole linked problem record was inaccessible here; neither the original argument nor a central AI disclosure could be checked.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The original Erdős problem record and the registry page could not be retrieved; claim scope and AI origin remain unverified.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The accessible primary note proves an extended-limit theorem and finiteness for subclasses, not the full finite-limit question; it contains no attributable central AI-origin account.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The original Erdős #709 record could not be retrieved, leaving the precise solution and central AI work unverified.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The cited Erdős problem page was unavailable, leaving the original AI and proof account unverified.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The Erdős page and linked solution archive were not readable here; the cited work appears to address partial certified bounds, so the exact claim and AI-origin step remain unresolved.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The original Erdős problem page returned 403 and no primary AI/proof disclosure was available for this entry.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The authors say ChatGPT helped verify and reformulate arguments and explore the structure across technical sections, but do not identify a specific model-originated chain-density theorem or proof idea.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The cited Erdős problem page was unavailable, so the original claim and AI role could not be checked.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The author reports extensive GPT-5.5 Pro exploration and proof development but also substantial correction; no specific retained model-origin idea is identified.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Discovery metadata attribution: What AI Found, CC BY 4.0 · VibeMathed contributors, editorial layer CC BY 4.0. AIProofWatch supplies the screening decisions.
Allow optional Google Analytics to help improve AIProofWatch? Privacy details.
