OpenAI

Greg Brockman’s journal really is making Greg Brockman look unreliable. Six days after telling Musk that Brockman et al. wanted more results in the nonprofit and to fundraise there, he writes, “We’ve been thinking about that maybe we should just flip to a for-profit. making money for us sounds great and all.” Brockman says this was an expression of a frustration and not a plan. Molo asks if he rehearsed that. Brockman says “no.”

He also writes, “To convert to a b-corp without him. That’d be pretty morally bankrupt.” That’s probably the most solid thing the Musk team has gotten out during the trial so far.

Molo asks if the nonprofit should still be a philanthropic endeavor. Brockman says, forcefully, “no.” We then look at notes from his diary, where he records Musk as saying “gotta figure out how do we transition from non-profit or something which is essentially philanthropic endeavor and is b-corp or c-corp.” Does make Brockman look pretty shifty.

We are hearing a lot of “I wouldn’t characterize it that way,” “I’m not sure I’d say it that way,” and “This sounds like something that I wrote, is it okay if we see it in context?” It’s not as contentious or tense as Musk’s exchanges with Savitt, but it’s definitely notable.

The question is whether Brockman disclosed his ownership of Cerebras when OpenAI was discussing merging with it. Brockman told Teller and Zilis, who were in theory Musk’s chiefs of staff, but not Musk himself. Molo is driving this home, which… I am less convinced by. Lots of executives delegate.

Musk’s team is finally landing some serious blows — namely that part of Brockman’s compensation was a grant from Altman’s family office. Musk’s adviser Jared Birchall wrote, “Greg is going to have a greater allegiance toward Sam as a result of this arrangement.” Brockman told Musk that “We ran out of YC stock fulfilling others’ offers” so Altman arranged the deal.

At a dinner about AI that Brockman and Altman attended, Musk was late. The first thing Brockman remembers Musk asking is, “Is Demis Hassabis evil?” Hassabis, of course, ran DeepMind.

It was initially supposed to be part of Y Combinator, as a research arm. Looking at a solicitation email from Brockman to then-CEO of Yahoo Marissa Mayer, Brockman writes that donors include Elon Musk, Jessica Livingston, Reid Hoffman, and Peter Thiel. “I’m personally donating $100,000,” he writes. But he didn’t end up donating that. Brockman is speaking very quickly and very softly, and YGR has just scolded him for it.

Brockman is taking the stand next.

I still don’t really understand how this expert helps Musk’s case, if at all.

We are now going through previous statements that Russell has made about Musk — for instance, Russell says that the “age of abundance” Musk talks about with AI robots will be great for Musk because of his pay package with Tesla. Also that Russell would not recommend that Musk be an AI adviser for President Trump.

The expert has testified in front of the US Senate about the dangers of open-sourcing AI systems. “If we open-source AI systems that are unsafe, we dramatically increase the risks,” Russell says. One of Musk’s contentions is that OpenAI is betraying its mission by not open-sourcing its models. Russell is now saying, in response to cross-examination, that open-sourcing can make it easier to remove safety guardrails that have been put in place. “It requires additional and very stringent safety measures.”

And he thinks posts are still called tweets. (The post is about AI creating so much wealth that “everyone can have a penthouse if they want.”) We’re getting some meandering analysis of AI creating prosperity, but even the judge seems frustrated at this point.

I really have no idea what it adds to Musk’s case. It seems to just be a way of running out the clock — but why bother? Is it just a way of giving OpenAI less time to defend themselves?

It’s very boring. Mainly we are establishing that the expert witness provided no specific opinions on OpenAI’s safety.

Because otherwise, this would be the guy to ask about the nonconsensual undressing and MechaHitler. We are getting slowly to the point, I think — which is that Russell suggests safety concerns would slow AI development. “Each company individually feels it needs to be in this race,” he says. “That means they can’t stop and solve the safety problem, which I think some of their employees would like to do, but the overall company police is preventing them.”

Russell is asked next about risk. He starts with individual risk, including algorithmic discrimination, which he says is “much more widespread” than people expect. He also mentions the possibility of AI systems reinforcing “delusional beliefs” — like, in his example, “Vladimir Putin is an alien.” Then there’s systemic risk — like AI systems taking jobs at a large scale. Russell mentions computer science students having trouble getting work already, and he expects it will accelerate.

His name is Stuart Russell, and he’s getting $4,000/hr for his first 40 hours of working with the team (and $1,500/hr after that). Seems like an expensive way for Musk’s lawyers to run out the clock… Because while we are getting some testimony about AI risks, I don’t really understand how this is relevant to the dispute at hand.

Russell is first asked about the positive aspects of AI — he mentions medical technology and AlphaFold, which can predict protein structures. Also mentions Waymo self-driving cars, which he describes as “extremely safe.” There can also be “broad economic benefits” where systems can be used to develop software, “where the AI can actually write software much faster and cheaper than human beings.”

Judge Yvonne Gonzalez Rogers says she got some “May the 4th be with you” jokes from her kids today as we get started. Apparently she appreciated them. After that, the day kicks off with expert testimony from computer scientist Stuart Russell.

We’re going through motions before the jury arrives. The motion I highlighted earlier this morning has been denied; YGR says the time to bring in the threats from Musk was when Musk was on the stand.

In response, Greg Brockman suggested both OpenAI and Musk drop their claims. Musk wasn’t interested. “By the end of this week, you and Sam [Altman] will be the most hated men in America,” he told them. OpenAI’s lawyers are trying to get the statement admitted to court as evidence that “Mr. Musk’s motivation in pursuing this lawsuit is to attack a competitor and its principals.”

In addition to our ongoing reporting from the courtroom and digging through newly released evidence, you will be able to listen in live via YouTube. While recording or rebroadcasting the stream is prohibited, we can expect it to be live generally between 11AM and 5PM ET while court is in session.

No courtroom updates today, but we have updated our rundown of the evidence with the latest exhibits added in Elon v. Musk, including details of the donated Tesla Model 3s.

Among the evidence released publicly, there’s this email exchange (Exhibit No. 844) between Valve cofounder Gabe Newell and Elon Musk about, of all things, trying to get a SpaceX tour and OpenAI introduction for Hideo Kojima.

After Birchall said he had no firsthand knowledge of the xAI bid for OpenAI’s assets, OpenAI’s lawyer asked that his testimony from the direct examination be struck. We are going to hear about that now, outside the presence of the jury.

He sees the OpenAI for-profit term sheet and writes to Shivon Zilis: “Pretty plain vanilla for-profit structure. So kinda hard to push a narrative that doesn’t involve investors being very focused on ROI. I’m a super fan of capitalism and making tons of money doing great things but not sure if this correlates with the ‘noble cause for humanity, not doing it to make money’ narrative. Did he/would he [Altman, lower on the email chain] offer E a board seat?”

We saw Chris Clark’s email about pausing donations yesterday during Musk’s testimony. Today, we see an email from Birchall: “This was ready to go out when I was told that Elon informed Greg and Ilya that the funding would be on pause until they came to terms on the right path moving forward.”
This was while there were discussions of how to put the for-profit Musk wanted together.

And we are talking with Birchall about tax deductions for charitable giving.

Users who enroll in the startup’s advanced account security settings can sign into their ChatGPT and Codex accounts using passkeys or physical security keys, and will also receive alerts about new logins to their accounts. OpenAI also automatically excludes users with the setting enabled from AI model training.

OpenAI and Musk’s counsel need to discuss something... Back in 15.

He was used, I think, to get financial documents into the record. We are now on the cross, and he is giving mercifully brief and direct answers.

There is a confirmation document from Chris Clark to Elon Musk showing the donation values. The donation was to OpenAI, as agreed in the stipulated facts, and in conflict to what Musk testified today.

We’re looking at a summary of about 60 donations to OpenAI, which Birchall says were directed by Musk, with Birchall helping execute all of them.

I guess she isn’t into Birchall’s testimony about Musk’s charitable contributions?