The Rapid Unscheduled Disassembly of OpenAI
The first salvo between those who want to accelerate the course of AI development as a force for good, and those who want to slow it down to better understand its effects and emergent properties.
To understand the history of OpenAI, Chamath Palihapitiya has an excellent article which help explain the corporate structure that led to Sam Altman's removal.
At roughly 4:00 PM on Friday Nov 17, OpenAI released a press release announcing a leadership change.
…Sam Altman will depart as CEO and leave the board of directors. Mira Murati, the company’s chief technology officer, will serve as interim CEO, effective immediately.
What’s interested about this release is:
It was put out at 4:00 PM on a Friday, a common PR strategy to try to bury the lead and miss the news cycle
Rather than phrase the transition for Sam and the company to save face (usual euphemisms include ‘he needed to spend more time with his family, or take time off to realign’,) the press release was specifically pointed and harsh.
On Friday, we were all expecting there to be claims of negligence, massive scandal, or breach of fiduciary duty. But as more details came out, the situation, or reason for Sam’s removal, became less clear. Not more.
Sam confirmed the news with this post on the X platform.
The board of directors at OpenAI included OpenAI chief scientist Ilya Sutskever, independent directors Quora CEO Adam D’Angelo, Co-Founder of Fellow Robotics and adjunct senior management scientist at RAND Corporation Tasha McCauley, and Georgetown Center for Security and Emerging Technology’s Helen Toner ( as well as co-founders Altman and Chairman/President Greg Brockman).
What’s interesting about this composition is that, since OpenAI’s governance structure is that of a non-profit (owning the for-profit OpenAI LLC subsidiary) most of these board members have no fiduciary responsibility to the investors of the company, which include Microsoft and Sequoia.
This was what was included in the official Release:
The board was not only ousting Altman, but also removing co-founder Greg Brockman as Chairman of the board, and promoting CTO Mira Murati to CEO.
Greg was to report to Mira, but the intention was for him to stay at the company. But late on Friday night, Greg quit the company altogether, effectively tying his reputation to Sam’s.
Later Friday night, Brian Chesky, co-founder of AirBnB and alum of Y Combinator (where Sam worked and was President for several years prior to OpenAI) posted on the X platform that “I have spoken to Sam and Greg and they will have a statement soon”.
What followed shortly was the statement by Greg:
What followed next was indeed an outpouring of support the likes I had never seen before. From people posting stories about the character and integrity of Sama, to condemnations of the board, to memes.
I kept waiting for clarity from the board, From Ilya. From anyone explaining their reasoning. None came. For whatever reason, they allowed Greg and Sam to shape the narrative, which for me, did not inspire confidence in the board’s decision making capacity.
The heavy hitters who had previously been on the board had long since left (Elon Musk, Reid Hoffman, Y Combinator founding partner Jessica Livingston; ex-PayPal CEO and Palantir Co-Founder Peter Thiel). The remaining members (with the exception of Sutskever), seemed drastically out of their depth.
What compounded this impression for me was that it came to light that none of the investors in OpenAI, including Microsoft or Sequoia, we told of the decision. They were informed one minute before the press release went out.
So Friday night as we all went to bed, we seemed to be still very much in the dark as to what had happened.
But Saturday morning, Nov 18, provided some clarity when an internal memo from the COO to the employees of OpenAI was leaked, confirmed Sam’s removal was not the result of malfeasance.
In response to this news, three of OpenAI’s top engineers quit.
Around the same time, it was leaked that Sam and Greg had already started making calls to raise money for a new AI company, and it seemed likely that these engineers, as well as many others, would defect OpenAI and move to this new company.
At the same time, concerned investors began a campaign to reinstate Altman and Brockman.
And from there the crazy saga continues. Around 5:30 PM EST we got this news.
And upon waking up on Sunday November 19, this was the state of the situation:
The former CEO, former Chairman and current CEO seemingly aligned.
Around midday on Sunday Nov 19, Andrej Karpathy, who served as the director of AI and autopilot for Tesla, and who was instrumental in creating and polishing GPT-4, posted this in response to a question about his silence:
Around that same time, PayPal alum, Craft Ventures founder, and All In bestie David Sacks posted the following on X, and was promptly responded to by Musk.
Late on the afternoon of Sunday Nov 19, due to investor and internal pressures, Sama was invited back to OpenAI to see if an accord could be struck for his return. Emily Chang reports:
Paul Graham, co-founder of Y Combinator and mentor to Sam, and Vinod Kholsa, early investor in OpenAI, made it clear where they stand:
Emily Chang reports an update close to 7:00 PM on Sunday Nov 19
Late that night ( Sunday Nov 19), the deal falls through. Sam will not be returning to OpenAI. The following deal is struck:
Huge W for Satya and Microsoft Shareholders, huge L for average OpenAI employees holding equity. Mira (who had been promoted to CEO) was replaced by Emmett Shear formerly of Twitch after she planned on rehiring Sam and Greg in her capacity as CEO.
What followed was an outpouring of support for Sam and a chilling reminder that companies are built and run by talented people, and without them…
But it STILL wasn’t over. Early Monday Nov 20, at around 8:15 AM EST, Ilya Sutskever, who has been instrumental in the coup (possibly the catalyst behind it) broke the silence he had maintained since the initial board announcement to issue an apology:
And then around 9:00 AM EST on Monday Nov 20, an internal memo was leaked which denounced the actions of the Board, including the ousting of both Sam and Mira Murati, and which included the language “you informed the leadership team that allowing the company to be destroyed ‘would be consistent with the mission of the company’”
The letter was signed by over 500 of the 700 OpenAI employees including Sutskever as well as CTO Mira Murati and COO Brad Lightcap.
Leading Paul Graham to weigh in once again:
I have no idea what the fallout of this will be, and if this drama has taught me anything, its useless to speculate.
But it feels like a hard fork in the timeline. The first salvo between those who want to accelerate the course of AI development as a force for good, and those who want to slow it down to better understand its effects and emergent properties.
The one question I still have. What did Ilya see what made him hit the panic button and kick off this whole fiasco in the first place?
Thanks for reading. Let me know your spicy takes in the comments.
UPDATE: Nov 20, 4:38 PM EST
It appears like Sam/Greg are still trying to return to OpenAI. Microsoft deal not done fully. Ilya has flipped his position to team Sam, joining Mira and the rest of the leadership team.
Employees appear to be in open rebellion:
The Independent Board Members appear to be absent:
Leadership team and Investors appear to be in alignment re: the path forward
UPDATED: Nov 21, 8:58 AM EST
OpenAI petition to remove the board reaches 743 employees, 95% of the employee base
Satya Nadella went on Bloomberg TV to discuss the situation:
OpenAI Board approached former employee and co-founder of competitor Anthropic about a merger. Anthropic is much more aligned with slow AI development than Microsoft, favoring slow and steady progress and closed source, Constitutional AI.
The situation is still dynamic with Sama working with the new CRO Emmett Shear to unify the company.
Meta quietly disassembles their own “Responsible AI” team. Meta’s Llama model is also open source.
We still don’t know the reason for Sam’s dismissal. Satya, internal employees, and other investors have yet to be informed.
Most likely reasoning is as fundamental rift between those who want to accelerate the course of AI development as a force for good, and those who want to slow it down to better understand its effects and emergent properties.
~This is a developing story and I’ll update it if it continues to progress ~
UPDATED: Nov 21, 4:01 PM EST
Emily Chang reports Sama and OpenAI board have been in talks all day. She now confirms if he were to return it would be as CEO only, not necessarily on the board.
Re: Board Member, Adam D’Angelo, CEO of Quora who appears to be the lynchpin after Ilya’s public apology.
Andrej Karpathy, who served as the director of AI and autopilot for Tesla, and who was instrumental in creating and polishing GPT-4:
OpenAI via the X Platform releasing the ChatGPT with voice feature, previously only available to premium members, in an effort to keep consumers on their app.
UPDATED: Nov 22, 9:01 AM EST
What a ride. It appears we have reached the conclusion of the saga (at least for now).
How did we get here? It seems like Helen Toner, Director of Strategy and Foundational Research Grants at Georgetown’s Center for Security and Emerging Technology was instrumental in ousting Altman.
Ms. Toner even went as far as to say “if the company is destroyed, that would be in line with the mission of creating safe AI.”
It appears a turning point in the negotiations was when interim CEO Emmett Shear asked for concrete proof of Altman’s wrongdoing. When none materialized, he threatened to resign, and began working as a bridge to bring Sam and Greg back.
A deal was struck in the late hours of the evening on Nov 21. Sam would return as CEO, Bret Taylor, co-creator of Google Maps while CTO of Facebook, chairman of Twitter, Inc.'s board prior to its acquisition by Elon Musk, and co-CEO of Salesforce, would join as Chairman of the board. Larry Summers, former US Secretary of the Treasury, would join as a board member, and Adam D’Angelo of Quora would remain. Ms. Toner and Tasha McCauley resigned as part of the negotiation.
Ms. Toner posted this on the X platform at 1:10 AM EST Nov 22:
Posts from all the key players:
This will no doubt be a case study reviewed in business schools for years to come, especially in regard to corporate governance. But I think its much more than that. When you look back at the 20th century, Yuval Noah Harari, the historian and author of Homo Deus, tracks how the conflicts of the 20th century were really about ideology. World War I was about the death of monarchy and empire. World War II was about the conflict between liberalism and evolutionary humanism (Eugenics, Social Darwinism and Nazism).
For me, this was the first salvo between Accelerists and Effective Altruists. Between the people who build, and the people who try to control. Between natural markets and central planning.
If you look back through the annals of history, central planning has never worked. Perhaps that will change through the advent of true AI, as a way to centralize and aggregate data.
But if this fiasco has taught me anything, it’s that people who pride themselves on their ability to think long term and protect us from AI, can’t even think through the second order consequences of their actions, let alone plan for them.
The end? I think not.