Skip to content
Moojaz — موجز
Search
Menu
HomeRegionsMaghrebAfricaMiddle EastEuropeAsiaAmericasWorldSectionsPoliticsEconomySecurityHealthScienceTechnologyAIEnvironmentCultureSportsSearch
File

OpenAI AI-generated scientific results and public examination

This dossier follows OpenAI's announced scientific results, proof availability and academic examination, distinguishing claims, formal checking and scientific understanding.

Last updated : 1 key moments
Corporate logo of OpenAI
OpenAI logo, used for identification. — OpenAI, Wikimedia Commons — public domain (PD-textlogo).

Where things stand

As of 28 October 2026, mathematics and computing materials are public; AGMAI requests release of physics proofs, with no established collective validation.

Reported information

OpenAI announces three scientific results as mathematicians request physics proofs

OpenAI announced model-generated results in mathematics, computer science and physics on 26 October 2026, with different review arrangements. The same day, the independent AGMAI group called for immediate publication of the physics materials withheld from public access.

Corporate logo of OpenAI
OpenAI logo, used for identification. — OpenAI, Wikimedia Commons — public domain (PD-textlogo).

The announcements concern harmonic analysis, a claimed disproof of the Exponential Time Hypothesis in computing, and the existence of certain quantum field theories. Mathematics and computing materials are public; OpenAI says a group of scientists is working to explain the physics materials.

AGMAI argues that examination should be open to everyone without privileged access. The announcements do not establish scientific consensus. The claimed Lean formal verification concerns encoded statements; it does not replace assessment of their significance. The claimed Yang–Mills existence result alone does not settle the full Millennium Prize problem, which also requires a mass gap.

Sources 4