Overview
- The Simons Foundation on Thursday launched a $15 million effort co-funded by OpenAI and Anthropic to help the math community assess and build on results produced by AI.
- OpenAI has pledged $5 million and both companies will provide model access and computing resources that Simons will distribute through an open call for proposals.
- The program will pay for workshops, conferences, special programs and formal-verification projects designed to check AI-produced proofs and teach researchers new tools.
- Independent human verification of OpenAI’s earlier release of hundreds of AI-generated manuscripts is still in progress and may take months because full prompts, model internals and training-data provenance have not been released.
- The initiative leans on Simons’s prior support for tools like the Lean proof assistant and aims to shift debate from quick public releases to community-driven standards, which could change how proofs are credited and vetted in mathematical research.