OpenAI 宣布其人工智慧系统生成了解决具有 200 年历史的纳维-斯托克斯方程（Navier-Stokes equation）的数学解，这项千禧年大奖难题的突破展示了 AI 推动高等数学的惊人潜力。为达成此成果，OpenAI 动用了多达上万个 AI Agent 并花费数百万美元的运算资源，最终在周末以 Lean 语言完成了形式化证明。

然而，这项成就随即引发学术归属与抢先争议。纽约大学数学家 Tristan Buckmaster 与 Anthropic 研究员 Levent Alpöge 指控，OpenAI 在获悉他们的相关进展后紧急加码资源抢先解题，并怀疑 OpenAI 检视了他们在 Codex 上的使用纪录，甚至在后续协商中试图排除 Alpöge 的挂名署名权。

面对指控，OpenAI 执行长 Sam Altman 及研究员 Sebastien Bubeck 坚决否认窥探其提示词或研究内容，强调双方解题方法在本质上截然不同，但也坦承无法完全排除去识别化产品使用数据对模型训练产生的间接影响。这场争端凸显出随著 AI 深入尖端科学研究，关于研究优先权与成果归属的矛盾将日益复杂。

OpenAI announced an AI-generated solution to the 200-year-old Navier-Stokes equation, a Millennium Prize problem, demonstrating the formidable capability of AI to advance pure mathematics. Achieving this milestone required deploying up to 10,000 AI agents and millions of dollars in computing power, culminating in a proof formalized in the Lean programming language over the weekend.

However, the breakthrough was overshadowed by controversy after NYU mathematician Tristan Buckmaster and Anthropic researcher Levent Alpöge alleged that OpenAI rushed to solve the problem after learning of their work. Buckmaster raised questions over whether OpenAI accessed their Codex usage logs and claimed OpenAI made proposals that would have excluded Alpöge's name from credit.

OpenAI leadership, including CEO Sam Altman and researcher Sebastien Bubeck, strongly denied accessing the researchers' prompts or proofs prior to public release, maintaining their mathematical approach was fundamentally distinct while acknowledging that de-identified telemetry cannot be entirely ruled out from model training. The clash signals growing tensions over priority and attribution as AI takes on larger roles in scientific discoveries.