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 results in mathematics, computer science and physics, and describes distinct plans for researchers to examine and explain them. The supplied report gives details for two results—a claimed proof of major conjectures in harmonic analysis and a refutation of the Exponential Time Hypothesis—but provides no substantive account of the physics result. Independent validation and the full supporting details remain central questions.
OpenAI says its models have produced three new results in mathematics, computer science and physics, and is using separate release plans to involve academic researchers in evaluating and explaining them. The company describes a claimed proof of the Exponential Time Hypothesis being false and a proof of major harmonic-analysis conjectures; it says the physics work concerns quantum field theory, but the supplied announcement gives few details about that result.
For the mathematics result, OpenAI says its models produced a proof of the local smoothing conjecture in every dimension. The company says this also resolves the Kakeya conjecture and Stein’s restriction conjecture in all dimensions. It attributes the proof to a novel auxiliary partial differential equation. OpenAI released a natural-language write-up and a Lean formal proof, a machine-checkable formalization, in its public GitHub repository. The company says the output was obtained on October 18, 2026, and formalized the next day.
The computer science result concerns the Exponential Time Hypothesis (ETH), a conjecture about the computational difficulty of Boolean formula satisfiability. OpenAI says its system proved the ETH false and that the result was subsequently verified in Lean. The company says an external group will help understand and explain the work, and that the group chose to make the natural-language proof and Lean formalization public immediately. OpenAI says the result could reopen questions about how efficiently related problems can be solved, but practical algorithmic improvements would require further research.
OpenAI also lists a result in quantum field theory. For that work, the external collaborators chose to hold back the AI-produced artifacts while they prepare their own write-up, according to the company. The announcement describes the release arrangements and broad areas, but the supplied material does not state the physics result itself or give enough detail to assess it.
How the Results Could Affect Research
The mathematics claims, if independently confirmed, would connect to several longstanding problems in analysis. OpenAI says the local smoothing result implies resolutions of the Kakeya and Stein restriction conjectures across all dimensions. Those claims concern fundamental questions about waves, focusing and geometric arrangements, and could influence later work in mathematical analysis. The announcement does not establish how the proofs will be received or what consequences researchers will draw from them.
The ETH claim has potential implications beyond theoretical computer science. The conjecture has been used as a basis for conditional hardness results in areas including graph theory, computational biology and databases, according to OpenAI. If the refutation holds, researchers may need to revisit results that rely on ETH and reconsider what algorithms might be possible. That does not mean existing problems have already become easier to solve in practice: OpenAI itself says finding faster algorithms and translating theory into practical gains will take substantial further work.
The announcement also tests a process for presenting AI-generated research. OpenAI describes the outputs as raw research data that needs expert interpretation, rather than finished scientific understanding. Its three release strategies give researchers different roles in reviewing, explaining and deciding what materials to publish.
Popular right now
Apple AirPods Pro 3 Wireless Earbuds with Active Noise Cancellation
Vitamix 5200 Professional-Grade Blender, 64 oz. Container, Black
De'Longhi Linea Classic Espresso Machine with Adjustable Milk Frother
As an affiliate, we earn on qualifying purchases.
A New Process for AI Research Claims
OpenAI says it consulted external experts, called “shepherds,” to organize groups around the results and advise on release. The approaches differ: the math output was released publicly for researchers to examine; the computer science work is being explained by an external group while its proof materials are public; and the physics group is preparing its own account before the AI artifacts are released.
The company says the collaborations will be transparent, voluntary and unpaid. Participants are not asked to sign nondisclosure agreements and may discuss work and publish conclusions. OpenAI says it does not seek academic credit for discoveries made by its models, while collaborators should receive recognition for their explanatory work and materials. It also acknowledges that choosing initial shepherds shapes who is involved first and creates information imbalances.
The announcement places the mathematics result alongside earlier developments: OpenAI notes that Guth, Wang and Zhang proved local smoothing in two spatial dimensions in 2019, while Wang and Zahl proved the three-dimensional Kakeya conjecture in 2025. Those prior results provide context, but they do not independently verify the company’s new, broader claims.
Independent Review Still Needed
The announcement reports that the computer science result was verified in Lean, and that the mathematics proof was formalized in Lean. Those are specific verification steps, but the supplied material does not provide an independent assessment of the arguments, the formalization’s scope, or the conclusions by outside researchers. OpenAI’s claims should therefore be distinguished from a broader academic consensus.
Key details about the physics result remain unavailable in the source material: its exact claim, evidence and expected publication or release schedule are not specified. The computer science section also ends before identifying the external working group or describing its review plan. The announcement does not say when the collaborations will publish their conclusions, or how disagreements about the results will be handled.
OpenAI acknowledges a further process concern: selecting initial shepherds can shape participation and leave others with less information. How the company will broaden access, and whether these collaboration models will be useful to researchers, remain open questions.
Researchers to Publish Their Assessments
OpenAI says it will provide materials and artifacts related to the results to the external collaborations. The computer science group has opted to publish the natural-language proof and Lean formalization immediately; its further explanation and assessment are still to come. For physics, collaborators are working on a write-up before the AI-produced artifacts are released, with no timetable stated in the supplied announcement.
For the mathematics result, the proof materials are already public, leaving researchers free to examine them and organize their own assessment. The next meaningful milestones are detailed write-ups, scrutiny by specialists and any corrections or responses that follow. OpenAI says it welcomes feedback on the release process, but has not specified a schedule for further updates.
Key Questions
What are the three areas covered by the announcement?
OpenAI identifies harmonic analysis in mathematics, computational complexity in computer science, and quantum field theory in physics. The supplied announcement describes the first two results more fully than the physics result.
What does OpenAI claim about the Exponential Time Hypothesis?
OpenAI says its system proved the ETH false and that the result was subsequently verified in Lean. The announcement does not establish that the claim has been accepted by the wider research community.
What has been released publicly?
OpenAI says it released a natural-language write-up and Lean proof for the mathematics result. For the computer science result, the external group chose to publish the natural-language proof and Lean formalization immediately.
What is known about the physics discovery?
The company identifies it as a result in quantum field theory. Its collaborators decided to prepare their own write-up before releasing the AI-produced artifacts, and the announcement does not specify the result’s exact claim.
Do the claims mean practical algorithms will improve now?
No immediate practical improvement is established. OpenAI says the ETH result could guide research into faster algorithms, but finding them and applying theoretical advances would require further work.
Source: openai.com
