TL;DR
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
OpenAI says its models produced three research results in mathematics, computer science and physics. It has published a formalized proof concerning harmonic analysis and an AI-produced result challenging the Exponential Time Hypothesis; the physics artifacts are being held back while an external group prepares a write-up. The claims, their verification and the details of the physics result remain subject to further scrutiny.
OpenAI says its models have produced three new results in harmonic analysis, computational complexity and quantum field theory, and is assigning different publication and expert-review processes to each. The company has released a natural-language write-up and Lean formalization for the mathematics result, as well as a proof package challenging a long-standing computer-science conjecture; it is withholding the physics artifacts while an external group prepares its own account.
The mathematics claim concerns the local smoothing conjecture, a problem about how waves focus over time. OpenAI says its models produced a proof in every dimension, which would also imply the Kakeya conjecture and Stein’s restriction conjecture in all dimensions. The company reports that it obtained the AI output on October 18, 2026, and formalized it in Lean the next day. It has published both an AI-generated natural-language proof and a Lean proof in a public GitHub repository.
For computer science, OpenAI says its system proved the Exponential Time Hypothesis (ETH) false, with the result subsequently checked in Lean. ETH is a widely used conjecture about the difficulty of Boolean formula satisfiability. OpenAI says an external collaboration will explain and examine the result, and that the group chose to release the natural-language proof and Lean formalization immediately. The supplied announcement does not identify the collaborators or provide a detailed account of the proof’s implications.
The third result concerns quantum field theory. OpenAI has not released the AI-produced artifacts, saying the external group working on the physics result chose to write it up itself first. The company describes the manuscripts as raw research data and says the aim is to help academic experts interpret and contextualize the results, rather than treating the outputs alone as finished scientific explanations.
Potential Effects on Complexity Research
If the ETH result withstands independent examination, it could affect a broad body of theoretical computer science. Researchers have used ETH as an assumption to establish conditional hardness results in fields including graph theory, computational biology, databases and computational geometry. A refutation would require scholars to revisit conclusions that depend on the conjecture and could alter how researchers assess possible algorithms.
That does not mean existing software will immediately become faster. OpenAI itself says that finding improved algorithms and turning theoretical results into practical gains would require substantial further work. The mathematics claim also has potentially wide reach: the company says its local-smoothing proof implies two other major conjectures. In both areas, the consequences depend on whether experts can validate the arguments and understand their scope.
The announcement also tests a process for AI-associated research. OpenAI says outside researchers will participate voluntarily and without pay, will not sign nondisclosure agreements, and will be free to discuss and publish their work. It says the groups—not the company—should receive recognition for the explanations and materials they create. These policies matter because the release strategy affects who can inspect a result and how credit is assigned.
Popular right now
Ninja Foodi 10-Quart DualZone 6-in-1 Air Fryer, Black, DZ302
Roborock Q7 M5+ Robot Vacuum and Mop
KitchenAid Artisan 5-Quart Tilt-Head Stand Mixer, Almond Cream
As an affiliate, we earn on qualifying purchases.
How the Three Results Differ
The mathematics problem has a long history. The local smoothing conjecture was proposed by Sogge in 1991. OpenAI notes that Guth, Wang and Zhang proved it in two spatial dimensions in 2019, while Wang and Zahl proved the three-dimensional Kakeya conjecture in 2025. The company’s reported result would extend local smoothing to every dimension, a much broader claim than those earlier results.
ETH is not a theorem but a conjecture about the time required to solve Boolean formula satisfiability. It has been widely believed and used to support conditional claims that certain computational tasks cannot be solved substantially faster. OpenAI says its new result was verified in Lean, a proof assistant that can check formal proofs against specified logical rules. Formalization can help detect gaps in an argument, but readers still need to understand what proposition was formalized and whether the proof’s assumptions and interpretation support the announced conclusion.
OpenAI says it asked external experts to shepherd the work and received different recommendations for each field: publish the mathematics output directly, organize an explanatory collaboration for computer science while making its proof public, and let the physics group lead the write-up before releasing artifacts. The company says it does not yet know which approach will work best and has invited feedback.
Independent Review Still Pending
The announcement does not establish that independent academic communities have accepted or published the results. Lean formalization is reported for the mathematics and ETH claims, but the supplied material does not include a full independent assessment of the proofs, their assumptions, or the interpretation of what they establish. The physics result is less accessible at this stage because its AI-produced artifacts have not been released.
Several details are also absent from the supplied report: the names and affiliations of the external experts, the full technical statements and proofs, and the identity or precise claim of the quantum-field-theory result. OpenAI acknowledges that choosing initial shepherds can shape participation and create unequal access to information. The scope of that concern and how the company plans to broaden involvement remain unresolved.
Proof Reviews and Physics Write-Up
The computer-science collaboration is expected to explain and examine the public natural-language proof and Lean formalization. OpenAI says the physics group will prepare a write-up before the AI-produced artifacts are released, but it has not given a publication date. The mathematics proof and its formalization are already available in the company’s public GitHub repository.
Readers should look for detailed technical papers, independent scrutiny and clear accounts of any assumptions or limitations. OpenAI says it will turn related materials and artifacts over to the collaborations and welcomes feedback on how to improve the process. Whether the three release strategies lead to broad participation, credible explanation and accepted results remains to be seen.
Key Questions
What are the three research areas?
OpenAI reports results in harmonic analysis, computational complexity theory and quantum field theory.
What does OpenAI claim about the Exponential Time Hypothesis?
It says its system proved ETH false and that the result was subsequently verified in Lean. The claim still needs detailed scrutiny and explanation by researchers.
Has the physics result been published?
Not in the materials described in the announcement. OpenAI says the external group chose to prepare its own write-up before releasing the AI-produced artifacts.
Does the ETH claim mean computers will become faster now?
No immediate practical improvement follows from the announcement alone. OpenAI says further research would be needed to find faster algorithms and translate theoretical advances into practical gains.
Source: openai.com
![Terence Tao: Math 2.0 [Pdf]](https://micronomicon.com/wp-content/uploads/2026/10/terence-tao-math-2-0-pdf-featured-560x300.jpg)
