OpenAI is trying to solve a problem that cannot be handled with an equation alone.
OpenAI announced Oct. 15 that it is contributing $5 million to a Simons Foundation initiative totaling $15 million in combined funding and resources from the foundation, Anthropic, and OpenAI. The announcement follows criticism of OpenAI’s release of hundreds of AI-generated mathematical manuscripts.
The program will fund researchers evaluating AI-generated work, learning new tools, and developing shared standards for using AI in mathematics. It gives the mathematical community more resources to deal with the technology, but questions about attribution, verification, and the sheer volume of AI-generated research remain.
Math community pushes back on OpenAI’s research rollout
The dispute intensified after OpenAI announced in September that an unreleased model had produced a proposed solution involving the Navier–Stokes equations. According to Euronews, two mathematicians who had been working toward a similar result questioned whether OpenAI’s work relied on theirs. OpenAI denied seeing their unpublished work.
On Oct. 6, OpenAI released a collection of mathematical manuscripts produced by an internal AI model. The Association for Human Mathematics responded that mathematicians “did not ask for this work to be done” and accused the company of disregarding established research norms.
The release also creates a substantial review task: mathematicians must assess correctness, novelty, and connections to existing research. OpenAI’s repository cautions that results are at different stages of verification and that some without formalized proofs could contain issues.
OpenAI says it has consulted an independent mathematics and AI advisory group, established protocols for revisions and citations, and published computer-checkable versions of many proofs using the Lean proof assistant. These steps provide ways to examine the work, but do not establish that every result is correct or resolve the attribution dispute.
OpenAI’s $5 million goes toward review and standards
OpenAI’s contribution is part of $15 million in combined funding and resources from the Simons Foundation, Anthropic, and OpenAI.
The Simons Foundation will manage and award the funding through initiatives that can include workshops and an open call for proposals. The goal is to help mathematicians evaluate AI-assisted results, learn new tools, exchange perspectives, and develop collective standards for AI use.
OpenAI and Anthropic will also provide AI tools and computing resources to researchers supported by the foundation. The resources can extend beyond mathematics into basic science.
Simons Foundation will administer the funding
The program’s structure puts the Simons Foundation in charge of administering the awards rather than leaving the process to the AI companies funding it.
According to Simons Foundation President David Spergel, the mathematical community should lead discussions over how AI tools are “evaluated, adopted and used,” rather than leaving those decisions to one company or institution.
The arrangement directly addresses one part of the criticism around who gets to set the rules for AI-assisted research. It does not, however, settle questions about how quickly AI-generated work should be released or how attribution should work when models produce results close to unpublished human research.
What eWeek found: The funding addresses review, not every complaint
Our reading of the announcements is that the initiative supports evaluation, access to AI tools, and community standards. It does not announce a resolution to the attribution dispute or a completed framework for publishing and verifying AI-generated mathematics.
Mathematicians’ concern | How the new program responds |
|---|---|
| Too much AI-generated work to review | Funding can support workshops and other programs where researchers examine recent AI-assisted results. |
| Researchers need access to the technology | OpenAI and Anthropic will provide AI tools and computing resources to researchers supported by the Simons Foundation. |
| AI companies could set the rules themselves | The Simons Foundation will manage the funding, while researchers will help shape collective standards for how AI is evaluated, adopted, and used. |
| Questions over attribution and unpublished work | The announced program does not establish new attribution rules or resolve the dispute surrounding OpenAI’s earlier mathematical work. |
OpenAI’s repository currently lists 719 manuscripts organized into 372 “families.” Each family groups related papers, which may include a principal result, supporting arguments, consequences, or alternative proofs. These figures describe the collection’s organization, rather than separate batches of discoveries.
The business implication is that review capacity belongs in the cost of adopting AI research tools. Producing more candidate findings can increase the work required to check them, even when the initial generation is fast.
For enterprises using AI in research, engineering, or other technical fields, a practical approach is to record the sources, assumptions, and model version behind each finding. Teams should also assign a qualified reviewer and define the evidence required before a result influences a design, investment, or published claim.
Verification and attribution require separate checks. A result may withstand technical scrutiny while still raising questions about whether earlier work received appropriate credit. Organizations need a process for both.
The Simons initiative may help the mathematics community develop shared standards. For businesses adopting similar tools, the immediate question is whether their ability to evaluate AI-generated findings can keep pace with their ability to produce them.
Want to learn more AI tips, tricks, and prompting techniques? eWeek readers get free 7-day access to The Neuron Academy, our practical learning platform designed to help professionals use AI more confidently at work. Browse all lessons →
Let us teach you How to Talk to AI for free! Try our six-minute course at The Neuron Academy and learn a few simple ways to write better prompts and get more useful results from AI, or browse our other AI course for free for seven days. Check out all the lessons here →