OpenAI 董事会风波

2023 年 11 月 17 日，一个震动整个科技行业的消息传开：OpenAI 董事会宣布解除 CEO Sam Altman 的职务。公告措辞模糊，只说是「审慎审查后认为他不够坦诚」。随后，联合创始人兼总裁 Greg Brockman 宣布辞职，首席科学家 Ilya Sutskever 成为焦点人物。全球最有价值的 AI 初创公司，在一夜之间失去了掌舵人。

接下来是戏剧性的五天。员工们迅速集结，超过 700 名（后来数字更高）员工签署联名信，要求董事会下台并让 Altman 回归，否则将集体跳槽到微软。微软 CEO 纳德拉公开表示欢迎 Altman 和 Brockman 加入微软，甚至传出了整支团队可能迁移的消息。与此同时，OpenAI 内部的人事、资金与客户关系都在剧烈摇晃。

11 月 20 日凌晨，微软宣布 Altman 将加入微软领导一个新的 AI 研究团队——看起来 OpenAI 的结局已定。但几小时后，剧情再次反转：谈判重新开启。11 月 22 日，OpenAI 宣布 Altman 回归 CEO 职位，董事会重组，Sutskever 退出董事会但保留公司职位。一场可能终结 OpenAI 的危机，在五天内以「创始人回归」收场。

这场风波留下的问题比答案多。为什么解雇？官方始终没有给出清晰解释，外界猜测涉及安全与商业化路线之争。谁真正控制 OpenAI？事件暴露了「非营利母体+营利实体」治理结构的脆弱——一个可以解雇 CEO 的董事会，在员工和资本面前几乎不堪一击。使命优先的治理设计，与快速膨胀的商业利益之间的张力，被原原本本摊在了全世界面前。

对产业的影响是深远的。OpenAI 的客户和开发者第一次意识到，自己依赖的 API 服务可能因一家公司内部的人事地震而中断，多模型备份开始成为企业标配。投资机构重新审视 AI 公司的治理结构。而 OpenAI 内部，Altman 复职后对公司治理进行了重新安排——那次地震虽然没有终结 OpenAI，却永久改变了它的权力结构。

回看这五天，它像一场提前上演的压力测试：当 AI 公司内部爆发治理危机时，员工、资本、竞争对手会怎样行动？答案已经很明确——员工用集体表态捍卫创始人，资本用橄榄枝施压，创始人用回归收尾。OpenAI 活了下来，但「AI 治理」这个词，从此不再是论文里的概念，而是一场全世界亲眼见证过的权力斗争。

On November 17, 2023, a message shook the entire tech industry: OpenAI's board had removed CEO Sam Altman. The announcement was vague, citing a "review process" that concluded he had "not been consistently candid." Soon after, co-founder and president Greg Brockman resigned, and chief scientist Ilya Sutskever became a focus. The world's most valuable AI startup had lost its leader overnight.

The next five days were dramatic. Employees organized fast, with more than 700 (the number later grew) signing an open letter demanding the board resign and Altman return, or they would defect en masse to Microsoft. Microsoft CEO Satya Nadella publicly welcomed Altman and Brockman, and reports even suggested the entire team might move. Meanwhile OpenAI's personnel, funding, and customer relationships shook violently.

Early on November 20 Microsoft announced Altman would join to lead a new AI research team—it looked like OpenAI's fate was sealed. But hours later the story reversed again: negotiations reopened. On November 22 OpenAI announced Altman would return as CEO, the board was restructured, and Sutskever left the board while keeping his company position. A crisis that might have ended OpenAI closed in five days with "founder returns."

The storm left more questions than answers. Why was he fired? No clear official explanation ever emerged; speculation centered on a dispute between safety and commercialization. Who truly controls OpenAI? The episode exposed the fragility of the "nonprofit parent plus for-profit arm" governance structure—a board that could fire a CEO proved nearly defenseless before employees and capital. The tension between mission-first governance and rapidly swelling commercial interest was laid bare before the world.

The industry impact was lasting. OpenAI's customers and developers realized for the first time that their API service could be interrupted by an internal personnel earthquake, and multi-model backup became enterprise standard practice. Investors re-examined AI companies' governance structures. And inside OpenAI, Altman reorganized the company's governance after his return—the earthquake did not end OpenAI, but it permanently changed its power structure.

Looking back at those five days, it reads like an early stress test: when a governance crisis erupts inside an AI company, how do employees, capital, and competitors act? The answer is now clear—employees rallied to defend the founder, capital applied pressure with an olive branch, and the founder closed the loop by returning. OpenAI survived, but the word "AI governance" was never again a paper concept—it had become a power struggle the whole world watched live.