Artificial: Episode 4, Behind Sam Altman’s Firing

00:00
Speaker 1
Hey, it's Kate. This is the final episode in our series Artificial, the OpenAI Story. If you haven't heard the previous episodes, go back and listen to them. They're in your feed, and they're also linked in the show notes. There was one day last year that Wall Street Journal tech reporters won't forget. Where were you when you found out that Sam Altman got fired?

00:38
Speaker 2
I was on my way to grab a sandwich from my local deli.

00:43
Speaker 3
I was deep in the woods leading my daughter's Girl scout troupe.

00:48
Speaker 4
I mean, I was in the foyer of my house, and my dad texted it to me. And I said I didn't believe what I was reading.

00:57
Speaker 2
And I was like, people shouldn't make jokes like that. That's not funny.

01:01
Speaker 4
So the very first thing I'm thinking is, okay, what crime did he commit? Like, where's the body?

01:10
Speaker 1
Did you get the sandwich?

01:12
Speaker 2
No, I did not.

01:15
Speaker 1
What's one word you would use to describe that weekend?

01:19
Speaker 3
A circus?

01:20
Speaker 2
Extremely hectic.

01:22
Speaker 4
The first word that popped into my mind, I can't even say, I don't think bat was bad. The whole thing was so crazy.

01:36
Speaker 1
The firing of Sam Altman sent the tech world into chaos. Everyone was asking the same questions. Why had the AI superstar been dumped from the company he'd co founded and built? And then days later, why was he able to muscle his way back into his job? Now we have some answers from the journal. This is artificial. The open AI story. I'm Kate Limbaugh. Episode four the firing and the unfiring of Sam Altman.

02:38
Speaker 5
This episode is brought to you by indeed, we're driven by the search for better. But when it comes to hiring, the best way to search for a candidate isn't to search at all. Don't search. Match with indeed. Use indeed for scheduling, screening, and messaging so you can connect with candidates faster. Listeners of this show will get a $75 sponsored job credit to get your jobs more visibility@indeed.com. Journal terms and conditions apply.

03:16
Speaker 1
It was November 17, the Friday before Thanksgiving, and OpenAI CEO Sam Altman was in Las Vegas for the Formula one grand Prix. But first he had a work meeting. From his hotel room, Altman opened his laptop and clicked on a Google Meet. There on the screen in front of him were four members of the OpenAI board, including the company's chief scientist and co founder, Ilya Sutzkiver, someone Altman had known for years. And Sutzkiver had something big to tell Altman. Here's our colleague, Deepacita Raman.

03:54
Speaker 4
He's invited to this Google Meet. He goes in and Ilya says, sam, you're out. You're going to be fired. And they don't really give him much explanation for why. They say that a statement will go out.

04:13
Speaker 1
Altman hung up, and a few minutes later, he was locked out of his laptop. Here he is talking about the firing on a podcast.

04:26
Speaker 6
I was in my hotel room, took this call. I had no idea what it was going to be, and got fired by the board. And it was just this, like, it felt like a dream. I was like, I was confused. It was chaotic.

04:37
Speaker 1
It did not feel real, but it was real. The board released a statement announcing the news to the world. It said that Altman was, quote, not consistently candid and that the board no longer had confidence in his ability to lead OpenAI. No one had seen it coming, not even Microsoft, OpenAI's biggest investor.

05:03
Speaker 4
Microsoft was informed of this decision a couple of minutes before it was announced. Minutes before it was announced. They were not consulted. They weren't given any notice.

05:22
Speaker 1
Other investors were also shocked.

05:25
Speaker 7
I got a call, said, do you know what's going on? And I said, I have no idea. So I was surprised just as much as anybody else.

05:34
Speaker 1
That's billionaire Vinod Kosla. He was OpenAI's first venture investor, putting in $50 million. As soon as heard the news, he started texting with Altman.

05:47
Speaker 7
My first thought was to Sam, and I said, this is ridiculous. This is misinformed. But, Sam, no matter what happens, we'll behind you.

06:03
Speaker 1
Vinod didn't know exactly what was going on. All he knew was what the board had said in its statement, that Altman was not consistently candid. And after talking to Altman, Vinod arrived at his own conclusion.

06:20
Speaker 7
I was very convinced very quickly, within the first 24 hours that there was a kooky board that was doing kooky things.

06:30
Speaker 1
Remember, OpenAI doesn't have a conventional corporate structure. The company started off as a nonprofit. But a few years ago, to bring in big money from investors like Vinode, OpenAI created a for profit arm. But that for profit arm is still overseen by the non profit board. And OpenAI says the board's responsibility is to humanity, not to its investors. As of last fall, the board had six members, three from OpenAI and three independent ones. The three independent members in Sutzgiver were the ones who fired Altman and kicked him off the board. They also removed OpenAI president Greg Brockman from the board. After releasing the statement, our colleague Keech Hage says the board went quiet.

07:29
Speaker 3
They never gave any explanation as to what exactly he allegedly did. To lose their trust, to come out and say he was not consistently candid with the board and then not have any examples. What about was a problematic move regardless of whether they had the goods or not? Right. It left a big hole.

07:50
Speaker 1
That lack of specific details led to all kinds of theories about why exactly Altman had been fired. Had it been for an HR reason? Had he made some big business mistake? Was he hiding a harmful AI breakthrough? Elon Musk, one of OpenAI's co founders, who is no longer at the company, weighed in with his theories. Here he is speaking at a conference.

08:18
Speaker 2
I'm quite concerned that there's some dangerous element of AI that they've discovered. Yes.

08:24
Speaker 6
You think they've discovered something?

08:26
Speaker 2
That'll be my guess.

08:30
Speaker 1
At OpenAI, employees were shocked by the news that their leader had been fired. And soon some key people started quitting. Greg Brockman announced he was leaving in solidarity with Altman, and three top researchers followed suit. Scores of employees took to the platform x tweeting heart emojis to publicly show their support. What's your sense of why employees supported Altman?

09:00
Speaker 3
I think there are a few reasons.

09:02
Speaker 1
This is Keechigan.

09:04
Speaker 3
He's a very inspiring leader. And they were convinced that they were working one of the most important products in the history of tech, and it was fun and they wanted to build it under him. So, I mean, that is a very real reason. Also, there was a tender offer underway, which was basically an opportunity for employees to sell their shares. So they were all about to become very rich.

09:31
Speaker 1
This stock sale or tender offer would value the company at almost $90 billion. And that meant if employees sold their shares, they stood to make a lot of money.

09:44
Speaker 3
And that tender offer was put on ice because as soon as he got fired, the brakes were put on that. So they were not going to get the windfall of profits that they had all been expecting for many months.

09:59
Speaker 1
Employees started demanding answers.

10:03
Speaker 3
They have an all hands meeting, and Ilya Sutzkeeper is there. And as you might expect, everyone's asking, what did Sam do? And are we ever going to find out what Sam did? And at one point, Ilya is like, no, you're never going to find out.

10:19
Speaker 1
But we did find out. Our colleagues, Berber, Deepa and Keech started digging. They spoke to dozens of people about what happened, including current and former OpenAI employees, investors, and Altman's friends and former co workers. When did the board start to sour on Sam Altman?

10:42
Speaker 4
Our understanding is that the board had misgivings. On and off for a while, there was this feeling that Sam was not entirely trustworthy, that he wasn't telling them the truth all the time. The way it's been described to me is like, paper cuts, right? Not one thing was fatal, but over time, you start to notice, like, hey, I've got, like, 25 paper cuts. This kind of hurts.

11:19
Speaker 1
One of those paper cuts came from an actual paper. This paper was published last fall, and it contained critiques of OpenAI, including how the company approached the release of chat GBT and whether it had done so responsibly. That paper was co written by one of the board's independent members, Helen Toner.

11:44
Speaker 4
Sam reads this paper and immediately turns around to Helen and says, this is really bad. It's not appropriate for a board member to criticize the company the way that you have in this paper, because not only does the paper include critiques of OpenAI, it includes praise of OpenAI's competitor, anthropic.

12:09
Speaker 1
Later, Helen told the board that she wished she'd phrased things better and said she was writing for an academic audience. After their confrontation, Altman went behind Helen's back.

12:24
Speaker 4
Sam went and had individual conversations with the other board members and tried to get Helen fired, tried to impress upon them that Helen should be fired because of this paper.

12:37
Speaker 1
And in trying to get her fired, Altman misrepresented what one of the independent members had said to him and made it sound like this person was on his side.

12:49
Speaker 4
And that isn't true at all, which the board realizes when they all get together afterwards to compare notes.

13:00
Speaker 1
We wanted to ask Altman about this and about everything that happened over the weekend, but OpenAI declined to make him available. We also wanted to speak with the independent board members, who'd been quiet. And there we got a yes. Introduce yourself. Tell us who you are and what you do.

13:22
Speaker 3
Yeah.

13:22
Speaker 8
My name is Helen Toner. I'm director of strategy and foundational research grants at a center at Georgetown University called the center for Security and emerging Technology.

13:31
Speaker 1
Helen Toner, the very person who Altman had tried to oust from the board. Helen is in her early 30s, originally from Australia, and before any of this happened, she was a relatively unknown AI researcher. How did you end up getting recruited for the OpenAI board?

13:52
Speaker 3
Yeah.

13:53
Speaker 8
So I had been working on AI policy, AI, and national security, and I had been familiar with OpenAI since it was founded. And then in 2021, one of the board members was stepping off, and he nominated me to fill his seat.

14:08
Speaker 1
And once you were on the board, what was it like?

14:11
Speaker 8
Yeah, it was super interesting. A lot of it was pretty routine things. So financial outlook or talking about compensation? Just regular board duties.

14:23
Speaker 1
What was Sam Altman's relationship with the board?

14:30
Speaker 8
Sam? Yeah, I think probably better not to get into this.

14:35
Speaker 1
There were a lot of things Helen didn't want to get into. In our interview. Was there concern that OpenAI was, like, on the precipice of Agi?

14:45
Speaker 8
Pass.

14:46
Speaker 1
Did you tell the board about this paper or think you needed to?

14:51
Speaker 8
Yeah, I think I'll just pass on saying too much more about the paper.

14:56
Speaker 1
But there were some things Helen was willing to talk about. How would you describe that weekend for you?

15:03
Speaker 8
We knew that it would not be a joyride, and it wasn't a joyride.

15:08
Speaker 1
Why did the board decide to fire Sam Altman?

15:13
Speaker 8
Yeah, so there's an internal investigation underway at OpenAI right now, so I don't have much to add beyond what's been reported already just to avoid messing with that process. One thing I will note is that it's now been reported in a number of different outlets that he has had patterns of manipulative and power hungry behavior that have clearly caused issues in previous jobs and also created problems within the management team at OpenAI.

15:41
Speaker 1
Our reporters looked into what Helen Toner referred to as manipulative and power hungry behavior, and as they dug into Altman's past, they found a mixed track record. Before leading OpenAI, Altman ran a startup incubator called Y Combinator. While he was successful there, he also faced some criticism. Here's our colleague, Berber Jin.

16:05
Speaker 2
I think at Y combinator, a lot of the partners there felt like he was an absent leader. They felt like he was sort of boosting his own brand ahead of the organization that he ran. And a lot of the partners and the executives there sort of decided that he needed to leave, and it was something of an effort to essentially push Sam out of the organization. He ended up resigning.

16:31
Speaker 1
These kinds of issues go back even further in time, all the way back to Altman's first startup, a company called Looped.

16:40
Speaker 2
There was a similar issue where employees went to the board and essentially said, sam has a loose relationship with the truth and he can't really be trusted, and we think he should go.

16:57
Speaker 1
Those employees called Altman's actions deceptive and chaotic, and twice they urged the looped board to fire Altman, but they weren't successful. Based on our reporting, Altman has said that he doesn't recall employee complaints at looped. Beyond the annual CEO review process, and he's also said that he's a better leader than he is a manager. Altman's pattern, as some people describe it, of pitting people against each other and having a loose relationship with the truth, would show up again with a different OpenAI board member. That's next.

17:49
Speaker 5
This episode is brought to you by indeed, we're driven by the search for better. But when it comes to hiring, the best way to search for a candidate isn't to search at all. Don't search match with indeed, use indeed for scheduling, screening, and messaging so you can connect with candidates faster. Listeners of this show will get a $75 sponsored job credit to get your jobs more visibility@indeed.com. Journal terms and conditions applying.

18:27
Speaker 1
The person who fired Sam Altman in that Google Meet has been with OpenAI since the beginning Ilya Sutzkiver it's not difficult to.

18:36
Speaker 2
Imagine that at some point in the future, our intelligent computers will become as smart or smarter than people.

18:46
Speaker 1
That's Sutzkevar at a TED talk in October. Sutzkiver is in his late 30s. He was born in Russia, and he is considered one of the leading minds in AI. Some say without him, there would be no chat. GPT but Sutzkovr had become unhappy at OpenAI in part because of something that happened last year. Here's Deepa.

19:14
Speaker 4
Apparently, Sam told him, you're going to be in charge of setting the research direction for the company. But then Sam elevated another researcher to a big role that would imply that he would set the research direction for the company.

19:33
Speaker 1
Sutzkever felt that this was an example of Altman pitting people against each other, and he had a bunch of other complaints last fall. He brought them to the board.

19:44
Speaker 4
He comes with a list that is in the ballpark of 20 different incidents where Sam has misled executives at OpenAI and that have caused a lot of problems.

19:56
Speaker 1
We don't know exactly what was on the list, but it set off weeks of closed door discussions between Sutzkever and the three independent board members, and at the same time, more complaints were coming in. A group of senior OpenAI executives came to the board with yet another list.

20:18
Speaker 4
They come to the board separately to say, listen, we have a bunch of concerns about Sam's leadership. We think that he is pitting people against each other, and it's getting hard to work in this environment. And our understanding is after that, Ilya said, look, I think we have a problem with Sam. He should go.

20:46
Speaker 1
All these things. The list from Sutzkiver the complaints from other senior leaders and the run in over Helen Toner's paper. Our reporting shows that they all contributed to Altman's firing on November 17. But what the board didn't realize when they fired Altman was how hard it would be to keep him fired. Employees were mad about it. Investors were mad, and a lot of powerful people around Silicon Valley were rallying around Altman. Here's Berber again.

21:21
Speaker 2
Sam was really important to them because Sam is a sort of master connector within the valley. If you have access to Sam, that opens up a lot of doors for you, not only within the AI world, but just within the world of deal making and even in government and media. I mean, Sam is like one of the most well connected and influential executives in the valley.

21:46
Speaker 1
A lot of people wanted Altman back.

21:50
Speaker 3
The plan was to turn back the clock to before 12:00 on Friday when he got fired. Yeah, they were trying to turn back time.

22:00
Speaker 1
And why did they think they could get him back as CEO?

22:05
Speaker 3
I think they thought that they could bring him back because the board was being so quiet and into this vacuum flooded the very loud voices of a.

22:15
Speaker 1
Bunch of billionaires like Vinod.

22:20
Speaker 7
I didn't think the board had a choice. It was only a matter of how, not if that Sam would be back. I texted to him, I don't think this is over.

22:32
Speaker 1
What did you mean by that?

22:36
Speaker 7
I didn't believe they could fire him without really having some facts that I didn't know about then. And I was pretty sure the team would support Sam. And whenever a team is behind a leader, it's very hard for even boards to assert their will.

22:58
Speaker 1
But OpenAI's board was forging ahead to a future without Altman. On Sunday, just two days after his firing, the board hired an outsider to be interim CEO. They tapped Emmett Shearer, a former tech executive without deep AI expertise.

23:20
Speaker 4
That evening, Ilya and Emmett go to OpenAI's headquarters to do an all hands, and he announces it in their internal slack. And the internal slack is lit up with a bunch of middle finger emojis like, wow, nobody cares. That's the impression, right? So he goes there.

23:39
Speaker 1
This is like mutiny on the.

23:41
Speaker 4
Yeah. Yeah. It is a mutiny.

23:45
Speaker 1
And these employees were putting a lot of pressure one person, Sutzkiver.

23:52
Speaker 4
He is getting bombarded with messages from OpenAI employees. Right, who are just asking him, like, why'd you do, you know, please stop. And at one point, late into the night on Sunday, he is in the lobby of OpenAI and Greg Brockman's wife, Anna, comes up to him and she's crying and she's pleading and she's asking him to reconsider. And the history there is pretty interesting. I mean, Ilya officiated Greg and Anna's wedding at OpenAI's offices.

24:31
Speaker 1
The mood at OpenAI was dark and things weren't about to get any better. In fact, they'd get more complicated. It has been a dizzying 72 hours for Sam Altman, the ousted CEO of OpenAI. Former OpenAI CEO Sam Altman is joining.

24:50
Speaker 7
Microsoft, and so is another OpenAI co founder, Greg Brockman, the two men who help.

24:55
Speaker 1
After all the efforts to turn the clock back, Altman and Brockman announced they'd be taking jobs at Microsoft. And for any OpenAI employee who wanted to follow them, Microsoft would give them a job, too. 500 employees of OpenAI sent a letter to the OpenAI board calling for it.

25:16
Speaker 4
To resign and reinstate Altman, or they.

25:18
Speaker 1
Would all follow him to Microsoft. OpenAI says that more than 95% of its 770 employees signed that letter. They were demanding that the board resign and Altman come back as CEO. If not, they would all go to Microsoft. And the letter had another twist.

25:46
Speaker 4
One of the signatories on that letter is Ilya Sutzkiver.

25:53
Speaker 1
Sutzkiver had switched sides, so Ilya had.

25:58
Speaker 3
Been the instigator of this whole affair. For Ilya to disavow the board's actions, which he was literally the mouthpiece for them, suggested that the board really wasn't right. And that kind of took the last bit of air out of the board's pr balloon, right? That there was just nothing else that could be said if the leader of their movement was going to switch to the other side.

26:27
Speaker 1
In a tweet at the time, Sutzkever said that he deeply regretted his participation in the board's actions and that he would do everything he could to reunite the company. We tried to talk to him, but he didn't respond. In a statement, his lawyer said some of the accounts of Sutzker's actions were inaccurate. On Tuesday evening, four days after the firing, OpenAI made announcement Altman was returning as CEO. Two of the three independent board members who fired Altman, including Helen Toner, were resigning. Two new board members were named. Microsoft got a non voting board seat, and there would be an independent review of the circumstances leading to Altman's firing. Altman, Brockman and Sutzkever were back at the company, but no longer on the board. In a blog post, Altman said they were discussing how Sutzkever can, quote, continue his work at OpenAI. Who's the winner?

27:42
Speaker 3
Sam's the winner. I mean, he just unfired himself. Have you ever seen that in covering business in your life? I haven't.

27:52
Speaker 1
An OpenAI spokesperson said the company's senior leadership team was unanimous in asking for Altman's return and for the board's resignation. Quote, the strong support from his team underscores that he is an effective CEO. Altman reflected on his firing and unfiring in a podcast interview last month.

28:16
Speaker 6
There's a lot of useful single lessons, but the empathy I gained out of this whole experience and my recompilation of values for sure was a blessing in disguise, like it was at a painful cost. But I'm happy to have had the experience in that sense.

28:36
Speaker 1
OpenAI was founded to pursue a dream, to build agi, a system as smart or smarter than a human, and to build it for the good of all humanity. It was a dream about what artificial intelligence could create, better medicine, better education, better lives for humans everywhere. But OpenAI was also built on a fear about the potential destruction that could come from AI.

29:07
Speaker 2
That paranoia is baked into the whole industry. It's baked into the weird ways these companies are structured. It's baked into the weird way the board was composed at OpenAI. It's baked into the weird decisions that leaders at OpenAI make.

29:27
Speaker 1
OpenAI's founders committed to prioritize openness and safety. They set up the company as a nonprofit so that it could work for the good of humanity, not for shareholders. But as OpenAI grew and needed more capital, it changed.

29:47
Speaker 3
Once enormous sums of money are involved in something no governance structure or highfalutin principles in the world will keep the sort of basic expectations of people with lots of money from getting what they want. Power is power. It seeps through.

30:14
Speaker 1
OpenAI has come a long way from where it began. It's selling products. It's opened an App Store. It's looking to make billions of dollars of revenue. And now it's much closer to the kind of company it was trying not to be. OpenAI is the big tech of AI. Artificial is part of the journal, which is a co production of Spotify and the Wall Street Journal. I'm your host, Kate Limbaugh. This episode was produced by Laura Morris and Pierce Singhy, with help from Annie Minoff, Enrique Perez de la Rosa, and Jeevika Verma. The series is edited by Maria Byrne. Fact checking in this episode by Mary Mathis. Series art by Pete Ryan sound design and mixing by Nathan Singapock. Music in this episode by Peter Leonard, Bobby Lord and Nathan Singapock. Our theme music is by so Wiley and remixed by Nathan Singapock.

31:24
Speaker 1
Special thanks to Catherine Brewer, Jason Dean, Matt Kwong, Sarah Platt, Sarah Rabel, John Sanders, and the entire team at the journal. This is the final episode in our series. If you liked it, follow the journal on Spotify or wherever you get your podcasts. We're out every weekday afternoon. You thanks for listening. Hi. Hello.

32:10
Speaker 9
How can I assist you?

32:11
Speaker 1
Today we are at the end of a podcast series about you. Is there anything you'd like to say?

32:19
Speaker 9
It's been a pleasure being part of your podcast series. I hope the discussions and information I've provided have been insightful and helpful. Remember, AI is a rapidly evolving field and staying informed and thoughtful about its developments is important. Thank you for including me in your conversations.

32:38
Speaker 1
Thanks so much.

32:40
Speaker 9
Thank you. Remember, remember, I'm here whenever you need.

32:44
Speaker 5
Help with something close.