Inside Sam Altman's firing: board dysfunction, Mira Murati's double role, and Ilya's pivotal evidence

Summary

Keach Hagey's upcoming book The Optimist: Sam Altman, OpenAI, and the Race to Invent the Future (May 20, 2025) provides a detailed account of OpenAI's November 2023 board crisis based on interviews with dozens of participants in Sam Altman's firing and reinstatement. The crisis exposed a fundamental structural problem: a nonprofit board with stated fiduciary duty to humanity controlled a for-profit AI company valued at $86 billion in a tender offer, with nearly 800 employees positioned to profit from equity sales.

Board power imbalance

Altman held no equity in OpenAI but exerted soft power over the board through sheer influence rather than legal authority. Board members Helen Toner (Georgetown Center for Security and Emerging Technology) and Tasha McCamey (former tech CEO, UK board member of Effective Ventures) grew frustrated that they lacked the technical depth to evaluate the actual risk profile of GPT models while Altman and co-founder Greg Brockman repeatedly blocked safety candidates the board wanted to add, including Aaya Kotra from Open Philanthropy. The board deadlocked for a year over which AI safety expert to hire.

Safety process failures

In winter 2022, Altman claimed three "controversial enhancements" to GPT-4 had been approved by OpenAI's joint safety board with Microsoft. When Toner asked for documentation, only one had actually received approval. Around the same time, Microsoft launched an unreleased GPT-4 test in India without formal safety board sign-off. Altman and Brockman made no mention of the breach in a subsequent six-hour board meeting. An independent board member learned about it only in a hallway conversation afterward.

The startup fund

In summer 2023, a board member overheard mention of OpenAI's startup fund at a dinner party. Directors later discovered Altman personally owned the fund, a venture vehicle the company had announced in 2021. OpenAI executives first claimed it was a temporary tax measure, then said it was simply faster to structure that way. Altman earned no fees or carried interest from the fund despite managing it, an arrangement some viewed as misaligned with venture norms but others interpreted as a legal structure ensuring all profits flowed through without personal enrichment to Altman.

Murati's account

Chief Technology Officer Mira Murati privately told board members that Altman's management style had created years of dysfunction. Brockman, technically reporting to Murati but with deep personal history with Altman from their Stripe days, would circumvent her authority by going directly to Altman. When Murati raised these issues with Altman directly, he responded by bringing OpenAI's head of HR into their one-on-one meetings for weeks until she clarified she did not intend to share feedback with the board.

Sutskever's break

Chief Scientist Ilya Sutskever told board member Toner that he had lost trust in Altman for multiple reasons, including Altman's tendency to pit senior employees against each other. Sutskever sent Toner self-destructing Gmail PDFs documenting what he characterized as Altman's lies and toxic behavior.

The broader context

Peter Thiel, an early OpenAI backer and former adviser to Y Combinator when Altman ran it, had repeatedly warned Altman that the company had been "taken over" by effective altruists concerned with AI safety. Thiel viewed these people as a threat to the company's mission. Altman deflected by noting that Elon Musk, another EA-aligned critic, had already been largely phased out. When the board finally moved against Altman, it was driven by some of the same safety concerns Thiel had flagged, though through a different lens and with far less clarity about what specific technical risks had triggered the action.