Skip to main content
Three Questions

What Went Wrong at OpenAI?

Days after he was suddenly fired, CEO Sam Altman returned to OpenAI on November 21, and the board that removed him was overhauled. We asked leadership expert Jeffrey Sonnenfeld how the company’s structure and the performance of the board contributed to the chaos.

Sam Altman

Sam Altman at the Asia-Pacific Economic Cooperation summit in San Francisco on November 16.

Photo: Justin Sullivan/Getty Images
  • Jeffrey A. Sonnenfeld
    Senior Associate Dean for Leadership Studies & Lester Crown Professor in the Practice of Management

What is unusual about the corporate structure of OpenAI?

It is very unusual to have for-profit subsidiaries of a nonprofit firm, but some organizations do—such as Underwriters Lab (UL), and Colonial Williamsburg and National Geographic in the past. In addition to confused objectives, the lack of clear accountability for directors to any constituency—investors, employees, customers, partners, regulators, etc.—is unusual.

Did that structure contribute to the removal of Sam Altman?

There was confusion about the obligation of the board to serve the financial objectives of its investors—as opposed to a primary obligation to serve society. Sam Altman managed to bridge different constituencies and schools of thought around AI, from the enthusiastic tech cheerleaders, to the confused commercial profiteers, to the industrial pragmatists, to the alarmed activists, to the global regulators. Investors and employees had no clear voice beyond threats to vote with their feet.

But there was also confusion as to who spoke for the anonymous board, a faceless group—and no accountability to explain their actions to anyone. To this day, no misconduct has been alleged and no evidence has been offered to support the original board’s loss of confidence. Meanwhile the board did not appreciate the uncommon loyalty Sam Altman enjoyed with all constituents— including its investors, its partners, its customers, and its 700 employees, who all threatened to quit and join Altman if he was not rehired and the board reconstituted. The trouble with this AI board was not its quality of artificial intelligence but its lack of genuine intelligence.

How would you rate the performance of the board?

The shadows here fall on the OpenAI board itself—clearly one of the most confused and incompetent in recent governance history with no apparent corruption or self-dealing involved. We never had names or clear accountable voices of this cowardly board throughout the five-day self-imposed leadership crisis, which imperiled the value of this firm, which had soared from $20 billion to almost $100 billion over just a few months. They never transparently revealed the reason for their loss of confidence in founder/CEO Altman with any misconduct or failures of judgment. There was no probation period, evidence of an outside leadership review for the board, or any warning to all key constituencies—who were all caught by surprise by the board ambush. They also massively miscalculated the sentiment of firm’s employee base—the data scientists who provide the genuine value of the company—and their incredible market demand. New board members Brett Taylor of Salesforce and former Harvard president Larry Summers are well-equipped to balance the humanistic, societal dangers against the capitalist commercial and technological frontiers of their nonprofit parent.

Department: Three Questions