OpenAI Abruptly Fires Sam Altman as CEO

On November 17, 2023, OpenAI's board of directors ousted Sam Altman as chief executive, in a decision announced without warning that has shaken the artificial intelligence industry. Altman, co-founder and public face of OpenAI since the launch of ChatGPT exactly a year ago, is also stepping down from the board.

A short, blunt statement

The board justified the move in a statement published on OpenAI's official blog, using a phrase now circulating across the industry: Altman "was not consistently candid in his communications with the board, hindering its ability to exercise its responsibilities." The statement offers no concrete examples and does not detail what this lack of transparency involved.

The statement adds that the board "no longer has confidence in his ability to continue leading OpenAI." There is no mention of any dispute related to products, model safety or business disagreements; for now, the official explanation is limited to this loss of confidence in his conduct before the governing body itself.

Mira Murati takes the helm on an interim basis

Mira Murati, until now OpenAI's chief technology officer, is stepping in as interim CEO while the organization searches for a permanent replacement. Murati has been with the company for several years and has been one of the public faces behind product launches such as ChatGPT and GPT-4.

The shakeup hasn't stopped at Altman's office. Greg Brockman, OpenAI co-founder and, until today, chairman of the board, has been removed from that role, though it was initially announced he would stay on at the company in another capacity. Shortly after the announcement, Brockman himself posted on social media that he had decided to leave OpenAI, without offering further explanation for his reasons.

Who is left running the board

Following the departures of Altman and Brockman, OpenAI's board is now made up of Ilya Sutskever, chief scientist and company co-founder, along with Adam D'Angelo, CEO of Quora, Tasha McCauley, and Helen Toner, an AI policy specialist. It is this small group that made and signed off on the decision to remove Altman.

OpenAI maintains a governance structure unusual among major tech companies: it operates under a non-profit foundation that controls a for-profit subsidiary, with the stated goal of ensuring that the development of artificial general intelligence benefits humanity above investor interests. That design gives the board far more power over the CEO than a conventional corporate board would have — and it is precisely the mechanism that made this ouster possible without shareholder approval.

A company at the height of its momentum

The news comes just eleven days after OpenAI held its first DevDay, the event at which Altman unveiled custom GPTs and GPT-4 Turbo to developers, with a speech focused on expanding the company's ecosystem. Weeks earlier, various outlets had reported on talks for an employee share sale that would value OpenAI at around $80 billion, a figure that would make it one of the most valuable startups in the world without being publicly traded. Announcement source.

Microsoft, OpenAI's lead investor and partner, with billions of dollars committed and deep integration of the company's technology across its products, has not yet issued a public comment on Altman's departure.

What we still don't know

The board has not said whether further changes are planned or how long Murati's interim term will last. It also remains unclear whether the decision stems from a specific, recent incident or an accumulation of internal friction. Based on information available as of this afternoon, Altman's ouster is being presented as an internal governance crisis, with the board not having tied the decision to any dispute over the safety or commercial direction of the company's models.

OpenAI has not yet called a press conference or responded to questions about Altman's future, whose next professional move remains, for now, unknown.

How to read a governance crisis without filling in the blanks

OpenAI's statement established only a limited set of facts: the board announced the departure, appointed Mira Murati on an interim basis and said it had lost confidence in Altman. It provided no examples with which to test the allegation. That absence is information in its own right. It permits reporting a breakdown of trust; it does not permit attributing that rupture to a product, a safety dispute or a particular negotiation.

The official November 17 statement also exposes an unusual structure. OpenAI, Inc., the nonprofit entity, exercised overall governance and controlled the commercial activities. Three kinds of power therefore need separating: legal authority to appoint or remove leaders; economic influence belonging to investors and partners; and operational power resting with executives and employees. They can point in different directions.

An investor can have enormous economic influence without holding the decisive vote. A board can possess formal authority and discover that using it destabilises the organisation. Employees may have no seat on the governing body yet change whether its decision is workable. Reading only the capital table or only the organisation chart leaves out part of the system.

A chronology protects against hindsight

During a fast crisis, preserve every document with its time, issuer and exact wording. Then separate confirmed facts, interested statements and open questions. If a later explanation emerges, it must not be inserted retroactively into an article dated earlier. Readers need to know what could be known at that moment, not a smooth narrative assembled with future information.

It also helps to note what did not change. The service continued operating, the statement announced no model withdrawal and did not connect the decision to a technical incident. Those absences do not prove that no other problem existed; they simply mark the boundary of public evidence then available.

The transferable skill is to distinguish formal authority, economic influence and operational legitimacy when a technology organisation enters a crisis. Before accepting an explanation, ask which document supports it, who had the power to decide and which part remains inference. Refusing to fill an institutional silence is a practical form of rigour.

Which document to request next

When a board uses general language to justify a dismissal, the next useful evidence is not rumour but governance material: bylaws, board composition, appointment powers, conflict policies and any minutes or review the entity chooses to publish. Each answers a different question. Bylaws show who can act; a later review may explain process; neither automatically turns a general allegation into a detailed proven fact.

Every claim should also retain attribution. “The board says” does not mean “it has been demonstrated”. “Brockman announced” identifies an interested source. A chronology must tolerate contradiction until documentary resolution arrives. That discipline prevents an organisation's confident tone from being confused with independent verification.

In companies combining a public mission with private capital, governance design deserves as much scrutiny as the product. The test is not whether the structure seems unusual, but whether responsibilities, controls and succession mechanisms were defined before the crisis and operate under pressure.

For readers, the quality signal is simple: every version of the account should lead back to the document available at that moment. If the document does not explain the cause, neither should the story. Temporal precision protects against immediate speculation and the later temptation to present an outcome as inevitable.