Technical‑Feature Correspondence between the OpenAI ETH‑Refutation Preprint and the "Multi‑Level Origin Dynamic Reincarnation" Framework, with Attribution Review

This preprint carries out a technical‑feature comparison and meta‑logical audit of the OpenAI preprint "The Exponential Time Hypothesis is False" (Oct 2026) which claims to refute the Exponential‑Time Hypothesis (ETH) for k‑CNF formulas. We compare its core algorithmic mechanisms: one‑parameter deformation, critical‑point splitting, polynomial reconstruction, bounded compression, recursive tensor‑network contraction, against our prior‑published framework "Multi‑Level Origin Dynamic Reincarnation". The prime‑number‑emergence sandbox manuscript was finished 2026‑08‑28 and publicly archived on Zenodo on 2026‑09‑01, fifty calendar‑days before the OpenAI manuscript release date of 2026‑10‑17. Functional‑structural isomorphism between the two iterative closed‑loop workflows is documented item‑by‑item via patent‑claim feature‑by‑feature matching. No definitive conclusion of plagiarism or patent infringement is drawn because internal access‑evidence remains unavailable. A central meta‑logical defect is analysed in‑depth: the OpenAI reasoning chain operates under potential‑infinity iterative‑evolution assumptions, yet jumps without justification to actual‑infinity complete‑object axioms when deriving universal conclusions for infinite families of k‑CNF instances. This creates an equivocation fallacy. The accompanying Lean4 formal‑proof scripts implicitly import an actual‑infinity completion axiom to make the script syntactically passable, constituting an ad‑hoc modification and begging‑the‑question fallacy. Lean4 only checks symbol‑level syntactic consistency and cannot detect meta‑level ontological conflicts between potential‑infinity and actual‑infinity. Public communication framing "Lean‑verified disproof of ETH" is misleading, since Lean validates a modified variant proposition with extra ad‑hoc axioms instead of the original Exponential‑Time Hypothesis. Leaving algebraic‑tensor calculus details aside, the paper argues that the overall ETH‑refutation argument in that OpenAI preprint suffers a foundational demonstrative fracture from this illicit ontological transition. This archive supplies reproducible feature‑matching tables, timeline records, and meta‑logical criticism for open peer review. It does not constitute legal opinion for patent or academic‑misconduct matters.