Deepening our collaboration with the U.S. Department of Ener…

科学进步塑造了从健康与能源到国家安全以及我们对宇宙的认知等方方面面。如果 AI 能帮助研究者拓展思路、更快检验假设，并将洞见更迅速地转化为经过验证的成果，其带来的好处将在学科间和整个社会中不断放大。

为此， OpenAI 与 U.S. Department of Energy 签署了一份谅解备忘录（ MOU ），以深化在人工智能与高级计算领域的合作，支持包括 Genesis Mission 在内的能源部倡议。这项工作属于 OpenAI for Science 计划的一部分——该计划旨在把前沿 AI 模型与真实科研环境中的工具、工作流和专业知识相结合，帮助科学家加速发现。

这份 MOU 建立在 OpenAI 与能源部国家实验室既有合作的基础上：我们已在真实科研环境中部署了前沿模型，并直接与科学家携手解决高影响力问题。

Genesis Mission 将政府、国家实验室与产业界聚集起来，利用先进的 AI 与计算加速科学发现。该 MOU 为信息共享与协调提供了框架，并为各方在具体项目成形时讨论和制定后续协议开辟了路径。与此同时， OpenAI 向 White House Office of Science and Technology Policy 提交了详细建议，说明美国如何通过 AI 强化科学与技术领导地位。这份文件阐明了我们为何把 2026 年视为“科学之年”，以及为何获取前沿 AI 模型、算力和真实研究环境对加速发现至关重要。与能源部的这项协议正是将这一愿景付诸实践的体现。

此公告发布之日，正值白宫举行的 Genesis Mission 活动， OpenAI 科学副总裁 Kevin Weil 与能源部及其他合作伙伴一同出席。

Kevin Weil 表示：“我们很高兴能与能源部合作，并为 Genesis Mission 做出贡献。当前沿的 AI 遇上国家实验室的专长，会打开新的探索方式，让想法更快得到检验并加速科学进展。”

共同加速发现的承诺

OpenAI 与能源部在推进基础与应用研究、强化美国在 AI 与先进计算领域领导力方面目标一致。该谅解备忘录为双方交换技术专长、协调活动并探索合作领域提供了结构化的途径——例如在聚变能源领域，能源部实验室拥有世界领先的设施、建模工具和数据——同时通过后续协议确保未来具体项目有明确范围与治理。

对 OpenAI 而言，这也体现了我们构建 OpenAI for Science 的思路：进展来自与研究者并肩工作，了解 AI 在何处能产生实质帮助、何处尚有不足以及如何安全地融入真实科研工作流。这样的工作依赖与领域专家的紧密互动、对一流科学基础设施的访问以及在科学实际发生的环境中进行严谨评估。该 MOU 为此类合作留出了空间，同时保留了负责任科学使用所需的严谨性和问责机制。

在与能源部国家实验室日益增长的合作之上

过去一年，我们与 DOE 国家实验室体系内的科学家密切合作，弄清楚前沿模型能在哪些方面发挥作用、在哪些方面不足，以及将其整合到真实研究场景需要什么条件。

把前沿模型交到科学家手中

我们与能源部国家实验室共同举办了 1,000 Scientist AI Jam Session——这是一次跨越九个实验室的首创活动，超过一千名科学家使用前沿 AI 模型来检验领域特定问题、评估模型回答并提供结构化反馈，推动未来系统开发。这种合作模式让研究者能在自己关心的问题上对 AI 进行压力测试，并共同塑造这些工具的发展方向。

在国家实验室超级计算机上部署前沿模型

我们与国土安全局核安全管理署（ NNSA ）下属实验室合作，包括 Los Alamos National Laboratory 、 Lawrence Livermore National Laboratory 和 Sandia National Laboratories ，支持科学与技术研究。作为这项工作的组成部分， OpenAI 已在 Los Alamos National Laboratory 的 Venado 超级计算机上部署了先进的推理模型，为 NNSA 各实验室的研究者提供共享资源。此项合作侧重于在高性能计算环境中应用前沿 AI 模型，以支持复杂的科学与国家研究挑战。

在现实情境中评估生物科学能力

我们还与 Los Alamos National Laboratory 合作，开发评估方法，研究多模态 AI 系统如何在实验室环境中被科学家安全使用。该工作旨在超越纯文本评估，朝着更贴近实际的测量方式迈进，以判断模型在高风险领域可能如何影响结果，并以专家监管、谨慎的研究设计和明确的风险降低承诺为基础。

OpenAI for Science 在此合作中的角色

OpenAI for Science 的目标就是加速科学：帮助研究者探索更多想法、更快检验假设，并发现否则需要大量时间才能得出的洞见。我们为实现这一目标持有两项互补信念：

• 科学工具至关重要。仿真引擎、分析流程、领域数据库和高性能计算对于精度与产能不可或缺。
• 前沿推理同样重要。随着模型体量和能力的提升，它们能在概念性工作中提供越来越多的支持：跨学科连接观点、在超出关键词检索的海量文献中导航、提出机制假说以及对假设进行压力测试。

能源部的国家实验室正好处在这两种要素的交汇点：它们运营着世界上最复杂的科学基础设施，并聚集了解决那些更好推理与更好计算能直接转化为科学进步与社会利益的问题的专家。

下一步

科学发现历来在优秀工具遇上优秀科学家时推进最快。我们相信，经与科学界紧密合作开发的前沿 AI ，可以成为一种新型的科学仪器：扩展研究者的探索范围、提升迭代速度，并帮助把洞见转化为实际影响。我们为能与能源部及国家实验室并肩工作、朝这一未来迈进而感到自豪。

Scientific progress shapes everything from health and energy to national security and our understanding of the universe. If AI can help researchers explore more ideas, test hypotheses faster, and move from insight to validated results more quickly, the benefits compound—across disciplines and across society.

OpenAI and the U.S. Department of Energy (DOE) have signed a memorandum of understanding (MOU) to deepen our collaboration on AI and advanced computing in support of DOE initiatives, including the Genesis Mission. This work is part of OpenAI for Science⁠, our effort to help scientists accelerate discovery by pairing frontier AI models with the tools, workflows, and expertise of real research environments.

This MOU builds on OpenAI’s existing work with DOE’s national laboratories, where we’ve already deployed frontier models in real research environments and worked directly with scientists on high-impact problems.

The Genesis Mission⁠ brings together government, national labs, and industry to apply advanced AI and computing to accelerate scientific discovery. The MOU establishes a framework for information sharing and coordination, and creates a path for the parties to discuss and develop potential follow-on agreements as specific projects take shape. Today, OpenAI also submitted detailed recommendations⁠ to the White House Office of Science and Technology Policy on how the United States can strengthen science and technology leadership through AI. That filing outlines why we see 2026 as a “Year of Science” and why access to frontier AI models, compute, and real research environments is essential to accelerating discovery. The agreement with the Department of Energy reflects that vision moving into practice.

This announcement follows today’s Genesis Mission event at the White House, where Kevin Weil, Vice President of OpenAI for Science, joined DOE and other partners. 

“We’re excited to collaborate with the Department of Energy and contribute to the Genesis Mission. When frontier AI meets the expertise of the national labs, it opens up new ways to explore ideas, test them faster, and accelerate scientific progress.”

A shared commitment to accelerate discovery

OpenAI and the Department of Energy share a focus on advancing basic and applied research and strengthening U.S. leadership in AI and advanced computing. This memorandum of understanding provides a structured way for OpenAI and DOE to exchange technical expertise, coordinate activities, and explore areas of collaboration—such as fusion energy, where DOE labs bring world-leading facilities, modeling tools, and data—while ensuring that future project work is clearly scoped and governed through follow-on agreements.

For OpenAI, this reflects how we think about building OpenAI for Science. Progress comes from working side-by-side with researchers to understand where AI meaningfully helps, where it falls short, and how it can be safely integrated into real scientific workflows. That work depends on close engagement with domain experts, access to world-class scientific infrastructure, and rigorous evaluation in the environments where science actually happens. The MOU creates space for that kind of collaboration, while preserving the rigor and accountability required for responsible scientific use.

Building on a growing partnership with DOE’s national laboratories

Over the past year, we’ve been working closely with scientists across the DOE national lab system to understand where frontier models help, where they fall short, and what it takes to integrate them into real research settings.

Together with DOE’s national labs, we convened the 1,000 Scientist AI Jam Session⁠—a first-of-its-kind event across nine labs where more than 1,000 scientists used frontier AI models to test domain-specific problems, evaluate model responses, and provide structured feedback to inform future system development. This model of collaboration allows researchers to stress-test AI on the problems that matter in their work, and help shape how these tools evolve. 

We partner with the National Nuclear Security Administration (NNSA) laboratories⁠, including Los Alamos National Laboratory, Lawrence Livermore National Laboratory, and Sandia National Laboratories, to support scientific and technical research. As part of this work, OpenAI has deployed advanced reasoning models on the Venado supercomputer at Los Alamos National Laboratory, where it serves as a shared resource for researchers across the NNSA labs. This collaboration focuses on applying frontier AI models in high-performance computing environments to support complex scientific and national research challenges.

We also partnered with Los Alamos National Laboratory⁠ to develop evaluations that study how multimodal AI systems can be used safely by scientists in laboratory settings. That work is designed to move beyond purely text-based assessments and toward more realistic measurements of how models may influence outcomes in high-consequence domains, grounded in expert oversight, careful study design, and a clear commitment to risk reduction.

How OpenAI for Science fits into this partnership

Our goal with OpenAI for Science is to accelerate science: to help researchers explore more ideas, test hypotheses faster, and uncover insights that would otherwise take significant time. We approach that goal with two complementary beliefs:

• Scientific tools matter. Simulation engines, analysis pipelines, domain databases, and high-performance computing are essential for precision and throughput.
• Frontier reasoning matters. As models scale and improve, they can increasingly support conceptual work, connecting ideas across fields, navigating large literatures beyond keyword search, proposing mechanisms, and stress-testing hypotheses.

DOE’s national labs are uniquely positioned at the intersection of these two beliefs: they operate some of the world’s most sophisticated scientific infrastructure and convene experts working on problems where better reasoning and better computing can translate directly into scientific advancement and societal benefit.

What’s next

Scientific discovery has always advanced fastest when great tools meet great scientists. We believe frontier AI—developed in close partnership with the scientific community—can become a new kind of scientific instrument: one that expands what researchers can explore, improves the speed of iteration, and helps translate insight into impact. We’re proud to work alongside DOE and the national labs as we build toward that future.

Generated by RSStT. The copyright belongs to the original author.

Source