OpenAI Releases Solutions to 372 Math Problems, Sparking Backlash

OpenAI has released solutions or partial solutions to 372 outstanding mathematical problems that have challenged human researchers for decades, according to a report by TechRadar. This massive publication effort, which spans 722 papers, marks a significant shift in the capabilities of artificial intelligence systems in advanced scientific fields.

The solutions were generated using an internal, unreleased AI model specialized for mathematics. While some of the provided answers are clear, others remain difficult for human mathematicians to understand or verify. TechRadar notes that the complexity of these solutions mirrors the difficulty of the original problems, creating a barrier to traditional scholarly verification.

Context and Background

Just two years ago, artificial intelligence systems struggled with GCSE-level mathematical challenges, according to TechRadar. The rapid advancement to solving decades-old problems highlights the accelerated development of AI reasoning capabilities. This release follows a proposal by OpenAI for the Navier-Stokes equation, which concerns the behavior of viscous fluids and was also generated by the company’s unreleased specialized math model.

The scale of the release has drawn criticism from prominent figures in the mathematical community. Renowned mathematician Terence Tao has called for colleagues working with OpenAI to abandon their projects. Tao shared a statement from the Association for Human Mathematics, which argues that releasing over 700 files at once demonstrates power rather than scholarship.

Who Is Affected

The Association for Human Mathematics urges mathematicians to return to a vision of science that centers human understanding. The group states that mathematicians did not request this work to be done. They also reference the Advisory Group on Mathematics and Artificial Intelligence, noting that the group’s initial statement advised frontier AI corporations against testing advanced mathematical problems on internal models.

TechRadar reports that the impact of this release has prompted a call for a return to traditional scientific methods. The controversy highlights a tension between the rapid output capabilities of AI and the traditional, slower pace of human-led mathematical discovery and verification.

What should you do?

Security and technology professionals should monitor how AI-generated mathematical proofs are integrated into software development and cryptographic systems. As AI models become more capable of solving complex problems, the verification of their outputs becomes a critical security concern. Organizations should review their reliance on AI for high-stakes mathematical computations and ensure that human experts can validate results before deployment.

What to do and how to stay safe: OpenAI

• Review internal policies regarding the use of AI for generating mathematical proofs or cryptographic algorithms.
• Ensure that any AI-generated mathematical solutions are independently verified by human experts before integration.
• Monitor industry standards and advisory groups for updated guidance on AI use in scientific research.
• Assess the transparency of AI models used in critical infrastructure projects to ensure accountability.

Step-by-step guide: Prompt Injection Warning Signs: How to Spot AI Manipulation

General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

How many math problems did OpenAI solve?

OpenAI released solutions or partial solutions to 372 outstanding mathematical problems, according to TechRadar.

Who criticized OpenAI's release?

Mathematician Terence Tao and the Association for Human Mathematics criticized the release, urging a halt to such collaborations.

Was the AI model used publicly available?

No, the solutions were produced using an internal, unreleased AI model specialized for mathematics.