Erdős Problem #896
On hold after reviewThe original problem page was inaccessible; the registry attributes an AI solve but supplies no readable original proof disclosure.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Considered but not curated.
Sources we considered, with review status and reasons.
- Pending: awaiting detailed review, not rejected.
- On hold: reviewed, with unresolved questions.
- Excluded or withdrawn: the recorded reason is shown.
- Related record: a linked source is already represented; companion claims may still await review.
Counts and coverage
3070 recorded entries · 652 awaiting detailed review · 584 excluded · 672 on hold · 3 publisher withdrawals · 1159 related to curated records.
- Includes recorded What AI Found, VibeMathed, OpenAI, and news-led primary-source entries, plus publisher withdrawals.
- Screening is metadata review, not a proof check. Registry flags are attributed reasons.
- Entries overlap; OpenAI entries group related papers. Counts are not distinct discoveries.
- Dates refer to saved decisions. Publication links use the current catalog.
- Recorded searches do not cover all research.
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.
Screening decisions
1911 matching entries · Page 17 of 39
Mixing Time of Kac's Walk on the Rotation Group
On hold after reviewThe paper proves an O(n² log n) mixing bound, but its accessible statement does not identify a central AI mathematical contribution.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #906
On hold after reviewThe draft gives a Gaussian construction but no AI disclosure, and the original problem record notes ambiguity about the intended transcendental version.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #888
On hold after reviewThe community entry attributes a proof of #888 to Aristotle and GPT-5.5 Pro, but the original Erdős record returned 403 and no first-person proof or model-origin account was available.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #1138
On hold after reviewThe source reports model involvement in developing the disproof but does not identify the central contradiction or construction supplied by a model.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-19T00:00:00.000Z
Erdős Problem #330
On hold after reviewThe linked original Erdős problem record was inaccessible, and the registry entry provides no original proof or model-work location.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Optimal Strategies in the All-Heads Coin Game
ExcludedThe authorship disclosure assigns the ideas, problem choice, perturbation plan and joint-induction structure to Pfaffelhuber; Claude executed exposition, code and Lean formalization.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #1190
On hold after reviewThe original Erdős problem record could not be read here; the community entry claims GPT-5.4 Pro and Lean verification but gives no original proof or first-person origin account.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #202
On hold after reviewThe original Erdős problem page returned 403; the queue does not identify a readable proof or AI disclosure for the asserted resolution.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #1014
On hold after reviewThe linked original problem record was inaccessible; the directory label alone does not establish the proof or AI role.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #863
On hold after reviewThe record credits Ho and GPT-5.4 Pro with noticing that existing literature plus a routine adaptation answers the question; it does not establish a new central AI-origin argument.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Sharp Convergence Rates for Viscous Hamilton-Jacobi Homogenization
On hold after reviewThe accessible Hamilton–Jacobi preprint provides the rate theorem, but no original disclosure of model-generated central mathematics was found.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #996
On hold after reviewThe cited lacunary-dilates preprint was readable but no original AI-origin disclosure was found, and the Erdős #996 record could not be checked. Missing keywords alone do not justify exclusion.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #603
On hold after reviewThe original problem page is a problem record rather than an accessible proof with AI provenance, so the specific model contribution cannot be identified.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #610
On hold after reviewThe community entry credits Aristotle and GPT-5.4 Pro with a Lean-checked graph-theory argument, but the original #610 record was inaccessible and the decisive AI step was not documented in readable primary evidence.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #522
On hold after reviewA new paper claims an almost-sure half-zeros result and Lean formalization, but its text gives no AI-origin account. Earlier forum AI claims concern a disputed stronger law, so neither establishes central AI work on this specific result.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Density Thresholds for Large Dilates of Point Configurations
On hold after reviewThe declaration says ChatGPT suggested approaches to Proposition 5, while the authors explicitly claim the main mathematical ideas and presented proofs; the retained model contribution to the density theorem is unclear.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Wickstead's Conjecture on Positive Projections
On hold after reviewClaude supplied details of Lemma 4.2, while the author claims all other results; the disclosure does not establish whether those details constitute the central result rather than completion of a human argument.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #858
On hold after reviewThe original Erdős problem page was inaccessible; the registry gives no first-party proof or model-origin passage for this exact solution.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #856
On hold after reviewThe linked original note proves an exact polylogarithmic exponent via weighted sunflower pressure, but its readable text gives no AI-origin disclosure. Hosting on ulam.ai does not establish model-authored mathematics.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #258
On hold after reviewThe community entry attributes a Lean-checked irrationality argument to GPT-5.4 Pro, but the original #258 record was inaccessible and the contribution could not be separated from any human work.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Exhaustive AdaBoost Cycling Question
On hold after reviewThe linked arXiv HTML could not be read, so the paper’s AI and proof disclosures remain unaudited.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #26
On hold after reviewThe community entry attributes a counterexample to Tenenbaum’s stronger variant to a DeepMind prover agent, while the original Erdős record was inaccessible; the formal-conjectures file is a statement, not proof provenance.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Bipartite Exact Matching in P
On hold after reviewGPT helped select theoretical routes and formulate equivalent reductions, while Claude implemented experiments and Aristotle formalized obligations; the original disclosure does not identify an AI-originated core algorithm or proof.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #1202
On hold after reviewThe sources credit a Price–GPT collaboration but do not identify who supplied the central interval construction; the human collaborator’s name also conflicts across sources.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-19T00:00:00.000Z
Erdős Problem #514
On hold after reviewThe linked original problem record was inaccessible, leaving the exact result and AI role unverified.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #380
On hold after reviewThe original Erdős problem page was inaccessible (403), so the reported solution and AI attribution remain unconfirmed.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Counting Partial Hadamard Matrices in the Cubic Regime
On hold after reviewThe disclosure describes model-guided analysis but does not identify a retained central model-origin step in the cubic-regime proof.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Sárközy's Conjecture on Sums and Products Modulo a Prime
ExcludedThe authors say Aristotle supplied Lean formalizations of four already established principal statements. The paper does not attribute the sums-products theorem or its proof strategy to AI.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #125
On hold after reviewThe original Erdős #125 record returned 403, so its proof and AI attribution remain unverified.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Improved Approximation Ratios for Multiway Cut
On hold after reviewThe paper credits ChatGPT with code for discovering rounding schemes and plots. It does not say whether AI proposed a retained scheme or merely implemented human-designed search. Curation treats an AI-found retained solution differently from code assistance.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #369
ExcludedSky Yang says the core CRT and cyclotomic construction was their own idea; GPT translated it into rigorous prose and helped find papers, while GPT and Claude later checked it. The described model work is editorial, search, and checking rather than central discovery.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The Lattice Triangle Problem in the Hard Obtuse Window
On hold after reviewThe revised theorem arose from human–model interaction, but the source does not identify a retained mathematical strengthening supplied by the model.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-19T00:00:00.000Z
Hamilton Decompositions of the Directed 3-Torus
On hold after reviewThe accessible Hamilton-decomposition manuscript does not establish which central return-map or odometer idea came from AI.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #1153
ExcludedTao credits himself with the decisive localization method; the model-assisted potential estimate is auxiliary to this final theorem.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-19T00:00:00.000Z
Ramsey-Style Hypergraph Partition Bound H(n)
On hold after reviewThe linked FrontierMath problem page supplies a question, but no accessible original solution or AI provenance was established.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #1148
On hold after reviewA first-person account credits GPT with helping repair an earlier argument, but the final proof does not identify which central step originated with the model.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-19T00:00:00.000Z
Pach's Tangency Conjecture: Improved Bounds
ExcludedThe paper credits Gemini only for back-and-forth on graph-theoretic Theorem 9; the tangency bounds in the queued claim are separate results with no documented central AI origin.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #650
On hold after reviewThe original Erdős problem page was inaccessible (403), leaving the solver, proof, and AI role unverified.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #457
On hold after reviewA secondary listing credits Aristotle and GPT-5.2 Pro with a Lean-checked solution of #457, but the original problem record could not be read and supplies no auditable AI-origin account here.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Uniform Witnesses for Uniform Set Systems: the k=3 Question
On hold after reviewThe cited Uniform Set Systems paper contains no AI disclosure; the linked ChatGPT share is titled as a witness-family construction but its conversation text was unavailable. The AI-origin result may be a later or narrower construction.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #966
On hold after reviewThe community record says Aristotle produced and formalized a construction for #966, but the original Erdős page was inaccessible; it also notes Spencer’s earlier unrecorded existence argument, requiring a careful novelty check.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #846
On hold after reviewThe secondary listing credits two independent AI efforts, but the original Erdős #846 record could not be read to verify either construction and its chronology.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
The Leading Constant for Large-Order Davenport-Schinzel Sequences
On hold after reviewThe original paper proves asymptotic tightness by optimizing the earlier S2 construction, but provides no account locating central model work.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
First Proof Question 8: Smoothing Polyhedral Lagrangians
On hold after reviewThe model’s proposed edge argument was replaced, and its proposed final criterion remains conjectural; a retained central AI result is not established.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #347
On hold after reviewThe secondary listing names several models and human collaborators for a Lean-checked solution of #347; the original Erdős page was inaccessible, so it is unclear who supplied the decisive complete-sequence argument.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Almost All Primes are Partially Regular
On hold after reviewThe paper is authored by Axiom Math researchers, but institutional affiliation alone does not establish a model-originated contribution; no specific AI mathematical role was found in the original text.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #1089
ExcludedThe model result was an independent rediscovery of an existing solution; the source makes no new result claim.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
A Generalization of Boppana's Entropy Inequality
On hold after reviewThe available account does not show whether a central model-generated step remains in the final theorem or establish the check scope for the claimed equality cases.
Screening recorded: 2026-10-17 · Decision reviewed: 2026-10-18T00:00:00.000Z
Erdős Problem #281
On hold after reviewThe Erdős #281 original problem record returned 403, leaving the claimed resolution and AI provenance unverified.
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.