Flip, flip, and flip again The story of the fight within OpenAI

Flipping, Flipping, and Flipping Again The Battle Within OpenAI

Author: Xi Xin, Hedgehog Community

Turn back the clock three days.

On the evening of November 16th Pacific Time, in a warehouse in Oakland, OpenAI CEO Sam Oatman was talking to a group of artists about artificial intelligence. Just a few hours ago, he delivered a speech on behalf of OpenAI at the APEC CEO Summit. This was one of the most important summits in the world in 2023, and as the most renowned tech prodigy in the world, he naturally couldn’t miss it.

It was a casual Thursday night, but with the arrival of a text message, the pleasant conversation had to come to an end. The sender was OpenAI’s Chief Scientist, Ilya Sutskever. It was a meeting notification message. Ilya, on behalf of the board, informed Sam to participate in a Google video conference the next day at noon.

• Apple Faces Class Action Lawsuit Over Blocking P2P Payments for Crypto Apps A Battle of the Tech Titans
• Apple in the Hot Seat Lawsuit Looms Over Blocked Crypto Tech for P2P Payments
• Major Banks Diving into Crypto: XRP Makes Waves

It seemed ordinary, considering he has been incredibly busy in the past year. With groundbreaking AI products like ChatGPT, OpenAI has become the most successful tech company of the moment. As the leader, Sam Oatman is in the spotlight and attracts a lot of attention.

No one could have imagined that he would face the biggest challenge since starting the company, and this challenge is not from competitors like Google, but from within OpenAI. Behind the text message, the “executioner” is ready, and an unexpected “internal struggle” is brewing.

OpenAI’s 72 Hours of “Palace Intrigue”

12:00 PM Pacific Time on November 17th (4:00 AM Beijing Time on November 18th): The Beginning

During the video conference on Friday afternoon, Sam Oatman faced the “trial” of the other four OpenAI directors. During the meeting, Ilya Sutskever announced to Sam Oatman that, with a vote of 4 to 2, the board had decided to remove him from his CEO position. Greg Brockman, the Chairman and President of OpenAI, who was also a staunch supporter of Oatman, was also “shown the door” without his knowledge or attendance at the meeting.

In the afternoon, OpenAI announced this news on its official website and also revealed the reasons for Oatman’s dismissal: “He has been consistently dishonest in his communication with the board, hindering their ability to fulfill their responsibilities.” The announcement also mentioned Greg’s resignation as Chairman of the board and announced Mira Murati as the interim CEO. As OpenAI’s CTO, she has been working with the leadership team for five years and is now taking on this role in a crisis.

The announcement quickly sparked global discussions. Shortly after the news was announced, Oatman posted on X, stating “I really enjoyed my time at OpenAI” and saying “to be continued.” A few hours later, Greg also posted a lengthy tweet, revealing more details. In just a short ten minutes, the fate of him and Oatman was decided by the four members of the board, and all of this happened “behind closed doors.”

Greg Brockman’s tweet

This “court intrigue” came suddenly, not only were both Ultraman and Greg kept in the dark, but most of OpenAI’s employees only found out about the change in ownership when the news was announced, which also sparked widespread dissatisfaction.

Pacific Time November 18th (Beijing Time November 19th): Reversal

Ultraman’s situation shocked everyone. He has always been the spokesperson for this company and GPT, and is regarded as a leader in the global AI industry. His dismissal by OpenAI came without any warning signs. Amidst conflicting rumors, conspiracy theories abound, and Ultraman’s private life and interim CEO Mira Murati became the center of attention.

In addition to OpenAI itself, Microsoft also became the hot topic.

Microsoft CEO Satya Nadella (left) and Sam Ultraman (right)

As the largest investor in OpenAI today, Microsoft had been gaining an edge in competition with rivals like Google thanks to the success of GPT. However, the surprise takeover of Ultraman undoubtedly threw everyone off balance. After the announcement, Microsoft’s stock price fell, and Microsoft spokesperson Frank Shaw released a statement expressing their intention to continue cooperating with OpenAI. At the same time, there were also reports that Microsoft had started working on Ultraman’s return.

The side effects of this “court intrigue” began to emerge. Within a day of the announcement, three senior researchers at OpenAI resigned, and internal employees began widespread protests; on the other hand, external shareholders like Thrive Capital started pressuring OpenAI, with various forces pushing the board to “revoke” their mistake.

By the afternoon of the 18th Pacific Time, there were reports that the OpenAI board had allowed Ultraman and Greg to return and had begun negotiations on this matter. In principle, board members were also willing to resign to make up for this “mistake”. However, they were far less decisive than they were a day ago when they dismissed Ultraman, and their hesitation on resignation and other issues further plunged the situation into confusion.

But Ultraman still feels an attachment to his colleagues at OpenAI. Shortly after the news of his return broke, he tweeted, “I really like the OpenAI team,” which was then retweeted by many OpenAI employees, accompanied by heart emojis. However, it is difficult to discern whether this is pressure on the OpenAI board or an expectation of Ultraman’s return.

Pacific Time November 19th 13:00 (Beijing Time November 20th 05:00): Negotiations

Two days after being fired, Sam Ultraman walked back into the familiar OpenAI building, but this time his purpose was negotiation. Before the negotiations began, Ultraman posted a photo of himself wearing a visitor’s badge on X, with the caption “This is the first and last time I will wear it [referring to the visitor’s badge],” representing two possibilities: either his return or never seeing him again. For him, this negotiation is more like a final ultimatum to the OpenAI board.

Things have developed to the point where everything seems to be in a delightful plot twist: the CEO, who was heartlessly fired, returns to the “throne” with the support of everyone and seeks revenge on the immature board members (by making them resign). It is certain that Ultraman negotiated with this mindset. According to reports from foreign sources, one of the important conditions for Ultraman’s return is for all four board members who fired him to resign. This has also become one of the main points of discussion between the two parties.

Surprisingly, interim CEO Mira Murali is one of the key driving forces behind Ultraman’s return. Shortly after the firing incident, she began negotiations with the board members in the hope of rehiring Ultraman and Greg at OpenAI.

However, the outcome of the negotiations seems less optimistic, with Mira even unable to keep her position as interim CEO. There are reports that, while negotiating Ultraman’s return, the OpenAI board is also looking for a new CEO, with former CEO of game streaming platform Twitch, Emmett Shear, apparently being their preferred candidate.

Emmett Shear, image source: Internet

It seems that everything is irreparable, despite constant efforts from shareholders like Microsoft to negotiate and employees threatening to quit. The OpenAI board members are determined not to back down, and Ultraman, naturally, cannot choose to surrender. The divide between them seems to be deep and insurmountable, as this tech unicorn faces an unprecedented rift.

November 19 at 12:00 AM Pacific Time (November 20 at 4:00 PM Beijing Time): The Decision

The plot twists once again. After the negotiations, the highly anticipated “Ultraman’s return” did not happen,

On the evening of the 19th, Ilya Sutskever announced that Ultraman would not be returning, and the new CEO would be Emmett Shear. Three CEOs in three days, OpenAI’s actions can only be described as abstract, rarely seen in the history of technological development.

On the midnight of the 19th, a tweet from Microsoft CEO Satya Nadella once again shocked all followers: Sam Ultraman and Greg Brockman will join Microsoft and lead a new advanced AI research team.

The plot becomes more abstract. According to previous disclosures, Microsoft has been actively pushing for Ultraman’s return, but the negotiations have clearly failed. In a situation where everything seems irreparable, Microsoft extends an olive branch to Ultraman, turning him from a co-founder of a startup company into the “client daddy.”

Ultraman and Greg retweeted Satya’s tweet with the caption “The mission continues.” In just one weekend, their careers underwent a drastic change, which was completely unexpected. But the shockwaves continue.

Interestingly, Musk became the biggest popcorn-eating audience in all of the twists and turns. Under Satya’s official announcement tweet, he jokingly commented, “Now they have to use Teams (Microsoft’s office software)!” (OpenAI’s board of directors fired Ultraman at a Google meeting, and Google is one of Microsoft’s biggest rivals)

November 20th 5am Pacific Time (November 20th 9pm Beijing Time): Departure

One call and everyone responded. If we were to describe Ultraman’s influence within OpenAI, these words couldn’t be more appropriate.

After the official announcement of joining Microsoft, the employees of OpenAI couldn’t sit still. In the early morning of the 20th, several employees, including Mira Murati, COO Brad Lightcap, CSO Jason Kwon, etc., started to speak up. They posted the same tweet on their respective accounts: “Without employees, OpenAI is nothing.” Ultraman retweeted each person’s tweet and added a heart emoji. This was a declaration of solidarity and also a “threat” to the OpenAI board of directors.

At this time, Ilya Sutskever seemed to realize the seriousness of the problem. He posted a message expressing regret for participating in the board’s decision, and stated that he would do everything possible to bring the company back together. Ultraman retweeted this tweet and added three heart emojis, whether it was sincerely forgiving or sarcastic is unknown. Netizens also speculated about Ultraman’s behavior: one heart represents that you can come with me to Microsoft, and three hearts mean you can’t. It’s like the Chinese social media expressions “dui” and “ah dui dui dui.”

A joint letter once again showed that the OpenAI board’s dominance was over. On the morning of the 20th, OpenAI employees issued a final ultimatum to the board members, threatening to collectively resign and follow Ultraman to Microsoft. The condition was that the entire board must resign and restore Ultraman and Greg’s positions. The letter stated that Microsoft had already assured the employees that positions would be available for those who joined, undoubtedly putting pressure on Microsoft under the breakdown of negotiations.

So far, 650 out of the 770 OpenAI employees have signed this joint letter, accounting for over 80 percent. Microsoft can easily absorb the entire OpenAI team without any bloodshed. Ironically, Ilya Sutskever’s name also appeared in the signatures, betraying the other board members.

Ultraman remains dignified. After receiving support from nearly all employees, he expressed that he was “extremely proud” and that “we are more united than ever before.” Although he lost in the internal power struggle, he gained recognition from everyone.

But he doesn’t want to see OpenAI completely torn apart. At 11 am Pacific Time on the 20th, he posted again, stating that his and Satya’s top priority is still to ensure the continued flourishing of OpenAI. There are reports that if board members are willing to step down, he and Greg are still willing to return to OpenAI.

Other technology companies who are watching with popcorn are already rubbing their hands together. Some have even released CEO email addresses, starting the open and legitimate road of poaching, in order to allow OpenAI employees who are prepared to resign to complete their “BOSS Direct Recruitment.”

In just three short days, OpenAI has undergone a complete transformation, from sudden dismissals to pressure for mass resignations. The board is facing unprecedented pressure, and OpenAI is facing an unprecedented crisis. But the game continues, and this ongoing drama of constant reversals seems far from reaching its conclusion.

Samuel Ultraman: The man OpenAI can’t do without

Many people wonder why Ultraman was so easily ousted? And why did he receive such great support?

The reason this “turf war” could happen is largely due to OpenAI’s unique organizational structure.

We need to review OpenAI’s history of establishment. OpenAI was founded in 2015 by Elon Musk, Samuel Ultraman, Reid Hoffman, and other Silicon Valley tech giants. At its inception, OpenAI was a nonprofit organization dedicated to “promoting and developing friendly artificial intelligence for the benefit of all of humanity.”

Realizing AGI (Artificial General Intelligence), benefiting all of humanity, open-source, nonprofit, and balancing Google AI – these were the reasons for OpenAI’s existence and its purpose. With a fervent passion for AI and $1 billion, OpenAI recruited the world’s top AI scientists. It was also at that time that Ilya Sutskever left Google and became the Chief Scientist of OpenAI.

From left to right: Ilya Sutskever, Samuel Ultraman, Mira Murali, Greg Robuckman

The nature of being a nonprofit organization also means that OpenAI will be different from typical tech companies. In other words, the OpenAI board’s main responsibility is not Microsoft and other shareholders, but “the whole of human society.” A significant part of measuring the company’s development is not based on profitability, but on “whether it promotes AGI for the benefit of humanity.”

It is because of this that the OpenAI board is not completely controlled by shareholders, so the decision to dismiss Ultraman was made without prior notice to external shareholders such as Microsoft.

Among the board members who voted in favor, in addition to Chief Scientist Ilya Sutskever, there are Adam D’Angelo, CEO of the American version of Quora, Tasha McCauley, a technology entrepreneur, and Helen Toner, Strategic Director of Georgetown’s Center for Security and Emerging Technology. Except for Ilya, the three directors do not hold positions at OpenAI.

The unique organizational structure and management model also led to a series of dramatic reversals: the board is not accountable to the shareholders, so despite the various pressures from Microsoft and other shareholders, they cannot control their will. From a certain perspective, the only thing that can balance them is “all employees”.

So why does Sam Ultraman have such influence?

As one of the co-founders of OpenAI, Ultraman is undoubtedly the soul of the entire organization. The technology team led by Ilya Sutskever created the revolutionary ChatGPT, and Ultraman is its biggest supporter.

Sam Ultraman’s biggest talent lies in business operations. He single-handedly completed the commercial transformation of OpenAI. Starting in 2018, Ultraman established a profitable division that could raise equity funds, driving OpenAI to move towards commercialization step by step. In 2019, under his leadership, OpenAI received a $1 billion investment from Microsoft and continued to sprint in a limited profit model.

After the official launch of GPT in 2022, with Ultraman’s efforts, the product gained outstanding results and matured in its commercial promotion. Once again, OpenAI received billions of dollars in investment from Microsoft, accelerating its growth.

Ultraman is a true pragmatist. Under his leadership, OpenAI has achieved rapid commercial transformation and continues to drive the implementation of AI applications. A memo from OpenAI disclosed on the 18th also showed that Ultraman’s dismissal had nothing to do with “malfeasance”, but rather with the breakdown in communication with the board. From this perspective, Ultraman’s work did not have any problems, and the main point of contention may lie in the future direction of the company’s development.

And from the series of reactions from the employees after the dismissal, Ultraman is very popular. It can be seen that he still has a considerable advantage in terms of downward management and individual capabilities. However, from another perspective, the employees’ “wholehearted assistance” is also related to the benefits that Ultraman can bring to them. As a radical proponent of commercialization, Ultraman is constantly turning OpenAI into a capital behemoth. For the relatively small-scale OpenAI team, he is bringing tangible “enormous wealth” to everyone.

The response from Microsoft and major shareholders is worth analyzing, for them, Sam Ultraman is undoubtedly the most suitable candidate for OpenAI. Microsoft is even willing to back him up, indicating his importance.

With such a “perfect” leader, nobody could have predicted that OpenAI, which had just released new products like GPT-4 Turbo and the GPT App Store, would encounter such a controversy. How did this crisis arise? Is it really a spur-of-the-moment “palace struggle”?

OpenAI, how did it come to this?

The answer may be hidden in the unveiling of GPT-4 Turbo.

In the article “How Many AI Startups Have Been Sentenced to Death by OpenAI’s Launch Event?” Hedgehog Society once discussed the dilemma that OpenAI is currently facing: prioritizing larger-scale commercial applications or pursuing technological breakthroughs for AGI?

From the beginning, compared to typical tech companies, the researchers and scientists in the OpenAI team have more influence, making them the most important part of the organization. This also led to OpenAI being like an ivory tower in the first few years of development, with the $1 billion investment not focused on profitability or expansion but solely on AI development.

However, as research deepened and more results were achieved, choices began to emerge. The first choice OpenAI faced was whether to transform into a commercial tech company.

This is the story of Musk leaving OpenAI. In that debate, Musk, who has always been cautious about AI development and still hopes to uphold open-source principles, ultimately failed to reach consensus with Ultraman and Iliad. In 2018, he chose to stop investing in OpenAI and stepped away.

In 2019, OpenAI officially announced its transformation, not only closing the source code but also making Microsoft its largest shareholder. Although the shift was from non-profit to a capped for-profit nature (with a profit cap of 100 times the investment), for OpenAI, developing AGI was no longer the only task at hand, and business and profitability were incorporated into their roadmap.

Sam Ultraman attending the APEC CEO Summit, image source: Internet

The subsequent story is well known. After three years of operation, ChatGPT emerged and OpenAI began to “conquer all sides”. In just one year, OpenAI not only accumulated 100 million users worldwide but also reached an estimated valuation of $90 billion. Apart from developing new large language model products, investing in startups, business operations, and diversifying profit models have become the top priorities in their current development.

The same choice has arisen again. In the rapid expansion of GPT, AI’s security issues are once again being taken seriously. On April 4, 2023, the Office of the Privacy Commissioner of Canada (OPC) announced an investigation into OpenAI, the company behind ChatGPT, involving allegations of “unauthorized collection, use, and disclosure of personal information by OpenAI”. Similar accusations continued to emerge in the following months, putting tremendous pressure on OpenAI regarding security concerns.

The scientists represented by Elya are also caught in a dilemma. On the one hand, GPT is continuously expanding its territory and enjoying the benefits of economies of scale, with research and development accelerating. With the increasing commercial value, OpenAI also has more funding and better research conditions, and the company’s profits are bound to take off. But on the other hand, AI faces the risk of going out of control, with endless security issues. As a technology company with the aim of benefiting humanity, OpenAI is becoming more and more aggressive.

Source: Internet

Especially at the Developer Day conference, the release of GPTs and the GPT app store directly pronounced the “death sentence” for many AI startups. Sam Ottman is radical and ambitious, and he certainly can lead the company to success, but everything seems to be deviating from the trajectory envisioned by scientists.

Perhaps it was under such disagreement that Elya launched this “attack”. It so happened that the other two external members of the board had a strong academic background and were also conservatives. With Elya’s pushing, they cast that crucial vote. Thanks to OpenAI’s unique organizational structure, Ottman was kicked out and the farce began.

Contrary to what many people imagined, this is not a case of capital cutting corners, but instead, Ottman is more like the spokesperson for capital. This is also the reason why Microsoft and other shareholders actively “rescue” OpenAI, as they profoundly understand that only under Ottman’s leadership can OpenAI unleash greater commercial value.

In the interim meeting after Ottman’s dismissal, Elya Sukskevich’s response seemed to indirectly confirm this point. In his view, this “palace struggle” stemmed from the board’s responsibility to fulfill the mission of a non-profit organization, which is to “ensure that OpenAI develops AGI for the benefit of all humanity”.

Fundamentally, this is a dispute over the company’s development path, and it is also a common issue faced by technology companies today: When technology and business collide, who should be the dominant force?

From this “palace struggle,” although Ottman was kicked out, it is undoubtedly Elya and the conservatives behind him who have truly lost power. They not only lost the support of the shareholders but also were “abandoned” by all the employees. Elya’s signature on the joint letter is undoubtedly a sign of weakness. This may be what he deserved. “Power games” and “attacks” cannot solve deep-seated divisions, especially when it comes to leaders with exceptional leadership qualities like Ottman.

But the biggest loser is still OpenAI.

As you can see, although they are playing a “life and death” game, the ultimate goal for both sides is to preserve OpenAI. Power struggles are certainly important, but avoiding the company’s division is even more crucial. As the joke goes, Microsoft is “painlessly acquiring” OpenAI, but even if Ultraman’s relationship with Microsoft is close, they surely don’t want their team to become a true “subsidiary” of Microsoft. For Microsoft, it is also not an easy task to re-incubate another OpenAI, especially with the looming problem of stock price fluctuations.

The story is far from over. A “palace intrigue” cannot resolve the route dispute under the influence of the business world on technological development. Perhaps in the future, this young company will be faced with many “choices” that will trouble them, which are the necessary pains to be experienced in the development of the AI field.

We will continue to update Blocking; if you have any questions or suggestions, please contact us!