As we reported this morning, Sam Altman has been reinstated as CEO of OpenAI and most of the remaining board has been removed and replaced. The dust is now settling on one of the most public corporate fallouts in decades and the picture about what exactly happened is clearer. Certainly, it now seems obvious who the person behind Altman's initial sacking was, and according to reports, that person is Helen Toner.
The Sam Altman Story: A Timeline of a Resurrected CEO
Before getting to the who, when, and why of the situation, perhaps it is worth breaking this whole saga down. Like a season of a TV show, this has been a flowing series with cliffhangers galore. If you have somehow missed everything – or even some – of what happened, here is the timeline:
- Friday 17, December: OpenAI announces that co-founder CEO Sam Altman has been removed from his position. Fellow co-founder Greg Brockman left with Altman, as did three key AI scientists.
- Saturday 18, December: There were already rumors that OpenAI's small board of six people was regretting its decision. Altman was reportedly back at OpenAI HQ negotiating a return to the company.
- Sunday 19, December: Microsoft CEO Satya Nadella confirmed the company's ongoing commitment to it OpenAI partnership. At the same time, he said that Altman, Brockman, and the three scientists would join Microsoft's advanced AI team.
- Monday 20, December: Ilya Sutskever, another OpenAI co-founder and apparent prime suspect behind Altman's ousting, expressed regret at the situation. Not only did this remove him as a suspect, but it also further opened the door for Altman's return.
- Tuesday 21, December: Nadella did live TV interviews suggesting that Altman could still return as CEO of OpenAI. Despite Nadella offering Altman a position at Microsoft, he said he was happy if he went back to OpenAI.
- Wednesday 22, December: Sam Altman is named OpenAI CEO again, with concessions such as most of the board being replaced and Altman and Brockman not taking a seat on the board. The announcement was welcomed by Nadella.
Of course, in between all of that has been some of the juiciest big tech and corporate gossip and rumors. Early reports pushed the idea that Altman had somehow unlocked artificial general intelligence (AGI) and was secretly developing a dangerous technology. Think The Terminator without the cool robots. Elon Musk, another (former) OpenAI co-founder and now competitor through x.AI, jumped on this. He urged Sutskever to disclose any AGI developments to the public. OpenAI's initial statement that seemed to cast Altman as dishonest only fuelled this speculation.
Why did you take such a drastic action?
If OpenAI is doing something potentially dangerous to humanity, the world needs to know.
— Elon Musk (@elonmusk) November 21, 2023
Who Has Been the Mastermind Behind the Saga?
As rumors of Altman masterminding a world-ending AGI eased, the fingers started pointing. Now the saga was a whodunnit, a corporate tale of back-stabbing. And his leads us to Helen Toner, who is now looking likely to be the driving force behind Altman's initial removal. Again, finding the prime suspect has been fluid over the last week, with a few people holding the title:
Ilya Sutskever: One of the people who along with Altman founded OpenAI. Sutskever initially supported Altman's removal and doubled down on the decision. However, by Monday he was backtracking and saying he had regrets. Sutskever was one of the 743 from 770 OpenAI employees who signed a petition letter to reinstate Altman. This seemed to remove him as the orchestrator of the whole situation.
I deeply regret my participation in the board's actions. I never intended to harm OpenAI. I love everything we've built together and I will do everything I can to reunite the company.
— Ilya Sutskever (@ilyasut) November 20, 2023
Adam D'Angelo: Over recent days, plenty of reports have suggested that Quora CEO and OpenAI board member, Adam D'Angelo, was behind the situation. I covered these reports in a detailed summary post yesterday. Essentially, the idea was OpenAI's recent GPTs service is a feature overlap of Quora's Poe AI service. In an effort to protect his own company, D'Angelo was rumoured to have tried to take Altman out and OpenAI down. His silence during the accusations only fuelled speculation. However, this seems to have not been the case as Adam D'Angelo remains on the new OpenAI board.
Helen Toner: OpenAI member and director of strategy at Georgetown's Center for Security and Emerging Technology, Helen Toner has now emerged as the likely candidate as the person who drove the corporate shambled at OpenAI over the last week.
Helen Toner: Last Person Standing
AI News YouTuber Wes Roth has summarized the growing rumour that Toner was the key board member in removing Altman. Before continuing, it is worth noting we are still within the realm of speculation. While some of the evidence looks compelling, there is no confirmation that Toner was behind anything. Even so, it seems there has been a growing rift between Toner and Altman for some time.
A New York Times article suggests the OpenAI board has been bickering for over a year. Amid the tensions, Toner co-wrote a paper with Georgetown that looked at openness in AI. It seemed the paper was critical of OpenAI's approach while being more forgiving of rival researcher Anthropic. Specifically, she praised Anthropic's restraint when releasing its Claude AI model, while arguing OpenAI cut corners when releasing ChatGPT. Altman is said to have personally met with Toner a few weeks ago to discuss the paper.
The Times points to a letter that Altman circulated internally within OpenAI that says he reprimanded Toner for the paper. Toner defended the paper as an academic report that looked at the challenges technology companies face with AI. According to the report, Altman was looking to remove Toner from the board and sought support from Sutskever.
Amongst AI founders, there is a surprising number who are concerned that AI could become a global threat. Former board member Elon Musk is one, and Sutskever is another. Instead of allying with Altman to remove Toner, he initially took the side of the board, helping to orchestrate last week's events.
It is worth noting that Anthropic was founded by Dario Amodei, the former OpenAI Vice President of Research. Amodei voiced concerns about Altman when at OpenAI, with reports suggesting he urged the board to remove the CEO. Amodei is said to have been one of the people OpenAI approached to replace Altman.
A Fractured Board: Altman vs. Toner
Following Altman's removal, the OpenAI executive team held a video call with the board, according to the New York Times report. They said Altman's dismissal put the company at risk. We already know that following the firing, OpenAI employees sided with Altman without question. However, Toner is said to have taken a hard stance during the call.
According to the report, she said the board's mission was to ensure OpenAI creates safe AI models that benefit “all of humanity.” She said if OpenAI was destroyed during the process of fulfilling this mission, then that would be ok.
The Wall Street Journal reports that during this call, the board was unable to give specific examples of Altman's dishonesty. One of the reasons initially given for removing Altman was that the CEO was not always candid or honest with the board. Babak Nivi, co-founder of startup funding firm AngelList went on X and pointed to the U.S. attorney's office in Manhattan calling OpenAI to seek reasons why the board is not being clear about Altman's dismissal.
The OpenAI board's hallucinations about Sam's lack of candor spurred law enforcement inquiries, according to WSJ.
This would be gross negligence and defamation.
On Friday, after the board defamed Sam in their blog post, the company got calls from law enforcement, including the… pic.twitter.com/HQ5kczOVp4
— Nivi (@nivi) November 22, 2023
The Sam Bankman-Fried Connection
Just when you thought this saga couldn't get any stranger, here we are. There is a small connection between Sam Bankman-Fried and Helen Toner and the OpenAI turmoil. Sam Bankman-Fried is a former cryptocurrency entrepreneur who founded FTX, a crypto exchange that collapsed into bankruptcy in 2022. He was convicted of fraud and conspiracy in 2023 and faces up to 100 years in prison.
He was known for his philanthropy and his support for effective altruism, a movement that aims to maximize the positive impact of charitable giving. The U.S. attorney office that contacted OpenAI is the same one that indicted Bankman-Fried. However, the links go further than this because Wes Roth points out that Helen Toner also seems to be a supporter of effective altruism.
Of course, there is nothing specifically wrong with that. However, it does make for an interesting aside on top of the saga.
Microsoft's Role in Saving OpenAI
Microsoft and its CEO Satya Nadella seemingly played a key role in reinstating Altman and saving OpenAI. When Altman left OpenAI on Friday, his suggestion of creating a new project was quickly ended when Satya Nadella snapped up Altman, Brockman, and AI scientists Szymon Sidor, Jakub Pachocki, and Aleksander Madry. Considering the mood amongst OpenAI employees, it is likely others would have followed. Salesforce was already making a public play for OpenAI's workers earlier this week.
Microsoft is spending at least $50bn annually on data centers going forward to power its AI push. @dylan522p notes this is “the largest infrastructure buildout that humanity has ever seen.”https://t.co/4GgWRb9Zw3 pic.twitter.com/DssJTyJMVl
— Edward Conard (@EdwardConard) November 20, 2023
What Microsoft was getting was the most prominent AI company completely for free, and without any regulatory oversight. Nadella and Microsoft were set to be the big winner from this. Of course, the company has already invested billions into OpenAI, but having the resources of the minds behind the company directly within Microsoft would have been a major advantage. It seems Nadella's actions – at least partly – forced OpenAI into making a swift turnaround and reinstating Altman.