The OpenAI Governance Crisis: What the Board Saga Means for AI Safety

The four-day sequence of events in November 2023 — Sam Altman's firing by the OpenAI board, a near-mass employee revolt, Microsoft's offer to hire the entire team, and Altman's reinstatement with a reconstituted board — has been exhaustively litigated in the press. What has received less systematic attention is what the episode reveals about the structural contradictions at the heart of OpenAI's unusual governance model. The company was incorporated as a nonprofit with a safety mission, then created a "capped profit" subsidiary to raise capital, then raised billions from Microsoft in a deal that gave the tech giant 49% of profits while the nonprofit retained nominal control. The board that fired Altman was exercising exactly the governance authority that structure was designed to provide; the market reaction demonstrated that real power had migrated to the commercial entity.

The key figure in understanding the fracture is Ilya Sutskever, OpenAI's chief scientist and co-founder, who voted to remove Altman before publicly expressing regret and signing the employee letter demanding reinstatement. Reporting from multiple outlets suggests Sutskever and other safety-oriented board members had concerns about Altman's communication style and the pace of commercialization, but could not articulate a specific trigger event that would justify the firing under the board's own standards. Sutskever's subsequent departure from OpenAI in May 2024 to found a new safety-focused lab, Safe Superintelligence Inc., reads as a tacit acknowledgment that the governance battle was lost: if you believe OpenAI's safety mission has been subordinated to commercial imperatives, you build a new institution rather than fight from within.

The aftermath reshaped OpenAI's governance in ways that concentrated rather than distributed power. The new board is smaller and includes more business-oriented members; the nonprofit's ability to independently oversee the commercial subsidiary has been further constrained. Altman has discussed converting OpenAI to a conventional for-profit structure — a move that would complete the transformation from safety-mission lab to AI company that began with the capped-profit structure. Critics argue this trajectory makes OpenAI increasingly indistinguishable from the commercial AI labs that its founding explicitly positioned it against. Supporters counter that frontier AI development requires the resources only a commercial entity can attract, and that safety research is better pursued from inside a well-funded frontier lab than outside it. Both arguments contain truth, which is precisely why the governance question will define AI industry politics for years to come.

Sam Altman returned as CEO on November 22, 2023. The reconstituted board includes Bret Taylor (chair), Larry Summers, and Adam D'Angelo.