OpenAI's AI Claims to Disprove Exponential Time Hypothesis and Solve Major Math/Physics Conjectures, Hands Verification to External Experts

OpenAI announced three new research results produced by its internal AI models, including a claimed disproof of the Exponential Time Hypothesis (ETH), a 1999 conjecture, and solutions to major math and physics problems, with verification handed to external expert groups. For broader context, explore our Top 100 AI Tools.

OpenAI's AI Challenges the Exponential Time Hypothesis

One of the most notable claims from OpenAI's recent research is that its AI system produced a proof demonstrating the falsity of the Exponential Time Hypothesis (ETH). The ETH, proposed in 1999, posits that certain computational problems cannot be solved in subexponential time. According to OpenAI, its AI discovered this disproof on October 16, 2026, and subsequently formalized it using the Lean proof assistant on October 17, 2026. For broader context, explore our AI Tools Pricing.

To ensure rigorous scrutiny, OpenAI has entrusted the verification of this ETH claim to a dedicated working group. This group is led by Russell Impagliazzo and Ramamohan Paturi, two of the original proposers of the Exponential Time Hypothesis, alongside MIT's Ryan Williams. Their involvement underscores OpenAI's commitment to having the findings reviewed by leading experts in the field.

Resolving Long-Standing Mathematical Conjectures

Beyond the ETH, OpenAI's AI models have reportedly achieved a significant breakthrough in harmonic analysis. The system generated a proof for Sogge's local smoothing conjecture, originally proposed in 1991. This proof is said to apply across all spatial dimensions and, in doing so, also resolves the Kakeya conjecture and Stein's restriction conjecture in all dimensions. These are complex problems that have challenged mathematicians for decades.

In line with its collaboration principles, OpenAI's math adviser panel (AGMAI) recommended that these results, particularly those formalized in Lean, be released to public scholarly repositories. OpenAI has followed this recommendation for two of the three announced results, providing access to the AI-generated proofs for broader academic review.

Advancements in Quantum Field Theory

The third major research result from OpenAI's AI involves a mathematically rigorous existence proof for interacting quantum field theories in four dimensions. This includes theories such as Quantum Chromodynamics and a unified completion of the Standard Model. Such a proof has profound implications for theoretical physics, potentially offering new insights into the fundamental forces and particles that govern the universe.

For this physics breakthrough, a working group led by Nobel laureate David Gross has taken on the task of writing up the result. Interestingly, this group has decided to withhold the raw AI artifacts for now, opting to present their own formalization of the discovery. This decision highlights the varying approaches external experts may take when verifying and presenting AI-generated scientific content.

OpenAI's Approach to AI-Generated Research Verification

OpenAI characterizes these AI outputs as "raw research data" and has established a unique verification process involving external "shepherd" groups. This collaborative model is designed to ensure the integrity and academic rigor of AI-produced discoveries. OpenAI's stated collaboration principles emphasize transparency, ensuring that AI discoveries do not receive academic credit themselves, but rather that the working groups and individual participants are recognized for their contributions.

This framework aims to foster independence among participants and ensure that the verification process is thorough and unbiased. By engaging original proposers of hypotheses and Nobel laureates, OpenAI is signaling a serious commitment to validating these potentially groundbreaking AI-generated scientific claims.

Key Takeaways for the Scientific Community

• OpenAI's AI has produced three significant research results, including a claimed disproof of the Exponential Time Hypothesis.
• The AI also reportedly solved Sogge's local smoothing conjecture, resolving the Kakeya and Stein's restriction conjectures.
• A mathematically rigorous existence proof for interacting quantum field theories in four dimensions was also generated.
• External expert groups, including original proposers and Nobel laureates, are verifying these AI-generated findings.
• OpenAI is releasing some AI-generated proofs to public scholarly repositories, emphasizing transparency.

Conclusion

OpenAI's announcement of its AI-generated research results, particularly the claimed disproof of the Exponential Time Hypothesis and solutions to major mathematical and physics conjectures, marks a pivotal moment in the intersection of artificial intelligence and fundamental science. The company's strategy of engaging prominent external experts for verification underscores a commitment to academic rigor and transparency. As these "raw research data" undergo scrutiny by the scientific community, the outcomes will not only validate or refute these specific claims but also shape future methodologies for AI-assisted scientific discovery. The scientific world will be closely watching the independent verification processes led by these distinguished groups.

Sources

• math/preprints/The-Exponential-Time-Hypothesis-is-False-October-25-2026/paper.pdf at main · openai/math · GitHub
• OpenAI announces three big science discoveries and a new way to work with academics | The Verge