The "palace intrigue" at OpenAI is still causing tremors.

TMTPost App news, November 22, Last night and early morning, around the ousting of former CEO of OpenAI, Ultraman (Sam Altman), three important events occurred involving the company's internal and external relations: disputes, a 4-hour collapse of ChatGPT, and Microsoft making a move to "poach" personnel.

First, according to multiple sources, a few weeks before his dismissal, Ultraman had a conflict with Helen Toner, the Strategic Director of the Center for Security and Emerging Technology at Georgetown and an independent member of the OpenAI Board. The main issue arose because Toner's team had co-written a paper that appeared to criticize the company’s safety methods and favored OpenAI competitor Anthropic. Ultraman tried to have Toner removed from the board and reached out to other members, but ultimately, it was Ultraman who was "ousted" by the entire board.

On the other hand, besides Ilya expressing "regret" for joining the dismissal, Adam D'Angelo, one of the core members of the dismissal incident and Quora CEO and OpenAI board member, still retains an opinion. He and the new acting CEO Emmett Shear represent one side, negotiating with Ultraman and the investor side. Bloomberg reported that the current round of negotiations is related to management chaos and employee threat of resignation, hoping to have Ultraman return as co-CEO, but the situation remains uncertain.

Secondly, a 4-hour crash of ChatGPT.

On the morning of November 22nd, Beijing time, ChatGPT and API developed by OpenAI experienced crashes and unavailability, lasting nearly 4 hours. As of the time of writing, ChatGPT and API have fully recovered. According to OpenAI, the main cause of the incident was due to issues with a database replica, affecting some endpoints of ChatGPT and API. Previously on November 8th, ChatGPT and API also experienced a major outage event that lasted over 12 hours, during which Ultraman tweeted an apology.

Finally, Microsoft's Chief Technical Officer (CTO) Kevin Scott posted on social platforms rallying: I saw the protest letter from OpenAI's employees, I appreciate it and invite you to join Microsoft’s new AI research lab led by Ultraman, and advance Microsoft's collective mission for AI development.

Earlier, after OpenAI’s board confirmed Ultraman's dismissal, Microsoft CEO Satya Nadella suddenly announced that Ultraman and former Open AI president Greg Brockman would join Microsoft to lead a new Microsoft AI team. Ultraman retweeted the post and stated: "The mission continues."

On November 21st, as the largest investor in OpenAI with an investment of $13 billion, Microsoft CEO Nadella expressed an "open attitude" towards Ultraman's next move, stating that whether Ultraman joins Microsoft or returns to OpenAI, he will collaborate with Microsoft.

"No matter where Sam (Ultraman) is, he is cooperating with Microsoft. I am still full of confidence in Sam, his leadership, and his capabilities, which is why we hope to welcome him to join Microsoft," Nadella said.

Reportedly, OpenAI has four independent centers of power: the Board of Directors; the senior management; the employees; and external investors led by Microsoft. The interests of the four parties are not entirely aligned.

(This article is first published on TMTPost App, Author | Lin Zhijia)

Below is the rough sequence of events regarding the Ultraman dismissal sorted by TMTPost App:

• On the evening of November 16th local time, former OpenAI CEO Ultraman and former president Brockman received a text message from Chief Scientist Ilya Sutskever, with Ultraman requesting a talk at noon on the 17th. They participated in a Google Meet meeting where, aside from Brockman, the entire board was present. Ilya informed Ultraman that he was going to be fired, with the news expected to break soon.
• At 12:19 PM local time on the 17th, Brockman received a text from Ilya asking for a quick callback.
• At 12:23 PM local time on the 17th, Ilya sent a Google Meet link. Brockman was informed that he was going to be removed from the board (although his position at the company was crucial and would be retained), and that Ultraman had been fired. Around the same time (4 AM Beijing time on the 18th), OpenAI published a blog post.
• The management team realized this shortly after, except for CTO Mira Murati who had found out the night before.
• The day before Ultraman was fired, at the APEC roundtable, he mentioned that next year, OpenAI would launch something more powerful than GPT-4 that is unimaginable and far beyond what people expect. "The capabilities of the model will leap in a way that no one anticipates. [It will be] different from what people expect. It will be amazing." Previously, Ilya repeatedly emphasized that ChatGPT might have become sentient, and AI will have an impact on humanity.
• On November 8th, ChatGPT and the API experienced their first significant outage, lasting over 12 hours, with Ultraman tweeting an apology at that time.

Beijing time, November 18th news:

• Local time on Friday, OpenAI announced that Ultraman would step down as CEO and leave the board, with Technical Director Mira taking over, effective immediately.
• OpenAI's board of directors tweeted that Ultraman's departure was a conclusion reached through deliberation by the board and due to his consistent lack of candor in communications with the board, which hindered the board’s ability to perform its duties, the board no longer trusted his ability to lead OpenAI.
• Ultraman stated: "I enjoyed my time at OpenAI, which was transformative for me personally, and I hope could also be transformative for the world. Most importantly, I enjoyed working with these talented individuals. More detailed things to say later on."
• Subsequently, OpenAI co-founder and president Greg Brockman also announced on social platforms that after resigning as chairman of the board, he decided to completely withdraw from the company and resign from all positions at OpenAI.

Beijing time, November 19th news:

• Ultraman informed investors of his plans to launch a new AI chip company aimed at building AI chips for tensor processing units (TPUs) to compete with Nvidia. Currently, his new company is in the development stage, Brockman may join it, and several investors including Sequoia Capital have expressed their willingness to support Ultraman.
• OpenAI's board of directors is discussing with Ultraman the possibility of him returning to the CEO position. He feels "conflicted" about returning and wishes to make significant changes to governance.
• Nadella has spoken with Ultraman to ensure that he could return to work at OpenAI on local time Monday.

Beijing time, November 20th news :

• Ultraman, as a "visitor" along with former president Brockman, met with executives at the company's San Francisco headquarters;
• It is rumored that Microsoft wishes to support Ultraman’s reinstatement and is considering placing an important figure in the new OpenAI board;
• At noon on the 20th, OpenAI announced the failure of Ultraman's return, and the board members selected former Twitch CEO Emmett Shear as the acting CEO. As Ultraman's departure was confirmed, dozens of OpenAI employees also chose to resign in follow-up.
• In the afternoon of the 20th, Microsoft CEO Nadella announced Ultraman, Brockman, and their colleagues would join Microsoft.
• Around 5:50 PM on the 20th, several employees including Mira tweeted in unison, "OpenAI is nothing without its employees," with Ultraman retweeting more than ten of these posts along with a "heart" emoji.
• On the evening of the 20th, 490 OpenAI employees collectively wrote a letter of protest, threatening the board with mass resignation unless the board stepped down and reinstated Ultraman as CEO. The letter stated: "The process of firing Ultraman and removing Brockman from the board has jeopardized all the company's work and undermined our mission. Your actions have clearly shown that you are not capable of overseeing OpenAI." Notably, the signatories of this letter include the company's director and chief scientist, Ilya Sutskever. This suggests that Ilya regretted joining the dismissal of Ultraman, reportedly changing his mind after talking to Brockman’s wife.
• On the night of the 20th, the new interim CEO Shear tweeted that his management team would investigate the board's dismissal of Ultraman, as the matter had affected the reputation of OpenAI.

Beijing time, November 21st news:

• The number of co-signers on the letter had exceeded 700 individuals, about 95% of the total employee count of approximately 770 at OpenAI. The letter also stated that the board must comply with all the employees’ terms, or the employees would collectively switch to Microsoft.
• Early on the 21st, Microsoft CEO Nadella stated to Bloomberg Television that Microsoft is "open" to Ultraman's next steps, whether he joins Microsoft or returns to OpenAI, Ultraman will collaborate with Microsoft. "No matter where Sam (Ultraman) is, he is working with Microsoft."
• According to The Information, aside from Amodio, other executives were contacted by the board to serve as interim CEO following Ultraman's dismissal, including GitHub CEO Nat Friedman and Scale AI CEO Alex Wang. Eventually, the board appointed Shear as the interim CEO on November 20th.
• Following the dismissal, over 100 OpenAI customers panicked over the event and reached out to competitors such as Anthropic, Google Cloud, and Cohere. It was revealed that many OpenAI customers were also considering switching to Microsoft Azure cloud services to provide OpenAI's GPT model technology. OpenAI had disclosed that there are currently over 2 million developers and customers developing on the ChatGPT API. Among Fortune 500 companies, 92% are using their products.
• The morning of the 21st brought news that the OpenAI board had discussed the possibility of a merger with competitor Anthropic, which is invested in by Google and Amazon, in the hope of replacing Ultraman with Anthropic’s co-founder and CEO, but the negotiation ultimately failed.
• In the evening of the 21st, Microsoft CTO Kevin Scott tweeted an invitation for OpenAI employees to join the Microsoft team, noting that he saw the OpenAI employee protest letter and invited them to join the Microsoft AI team.

Beijing time, November 22nd news:

• According to The Information, a few weeks before his dismissal, Ultraman had a conflict with Helen Toner, a strategic director at the Center for Security and Emerging Technology in the United States and an independent member of the OpenAI board, mainly because Toner's team co-authored a paper that seemed critical of the company's safety methods and favored OpenAI competitor Anthropic. Ultraman attempted to have Toner leave the board.
• Several early OpenAI employees claimed they have received invitations to join competitor companies such as Google and Anthropic. Meanwhile, the majority of employees asserted that they would resign if Ultraman was not reinstated.
• Anna Makanju, Vice President of Global Affairs at OpenAI, sent a memo to employees stating that the company has been in “intense discussions” with the board, Altman, and Shear to achieve unity in the company.
• Early in the morning of November 22nd, around 5 AM, ChatGPT and API developed by OpenAI again experienced outages and unavailability, lasting close to 4 hours. At the time this report was written, ChatGPT and API had fully recovered. OpenAI stated that the main cause of the incident was an issue with a database replica, affecting some endpoints of ChatGPT and API.
• At present, while Ultraman is still discussing with other executives and key shareholders the possibility of his return to OpenAI, the details of the disagreement have surfaced. If Ultraman returns, it is expected that the board of directors and its structure will undergo changes, and Microsoft may obtain a board position in some capacity, but the development of the entire situation still remains uncertain.