🤚 The Open-Palm Deposition
The Musk v. OpenAI trial — the tech industry’s most expensive divorce proceeding — continued this week at the Ronald V. Dellums Federal Building in Oakland, California, and Sam Altman took the stand to deliver testimony that was equal parts corporate history lesson and family therapy session.
According to Altman’s testimony, Elon Musk once contemplated transferring control of OpenAI to his children. Not selling. Not licensing. Not setting up a governance committee. Handing the organization that may eventually build artificial general intelligence to his offspring, like a particularly consequential trust fund or a horse ranch in Montana with existential risk implications.
The broader context: Musk is suing OpenAI and Altman, claiming they deceived him during the company’s transition from a non-profit research lab into a capped-profit entity that is now worth north of $300 billion. Musk’s position is that he was promised OpenAI would remain a non-profit dedicated to the safe development of AI for all of humanity. Altman’s position is that Musk wanted control, and when he didn’t get it, he left and started xAI instead — which, as we reported yesterday, has since lost 75% of its founding team.
👐 The Two-Handed Cross-Examination
Altman testified that Musk’s focus on controlling the for-profit entity concerned him specifically because “OpenAI was dedicated to keeping advanced AI out of the hands of a single person.” Drawing from his Y Combinator experience, Altman noted that he had learned founders who gained control “usually did not give it up.” This is, of course, a polite way of saying: I’ve seen what happens when billionaires get what they want, and it involves a lot of renaming things after themselves and firing everyone who disagrees.
The children angle is the kind of detail that lawyers live for and historians will spend decades contextualizing. The man who has publicly warned that AI poses an existential threat to civilization — who has compared it to summoning a demon — allegedly wanted to hand the keys to the demon-summoning organization to his kids. This is not a contradiction if you understand that, in the Musk cinematic universe, existential threats are just assets that haven’t been properly managed yet.
It also raises a question that no one in the courtroom was equipped to answer: which children? Musk has, at last public count, somewhere between 11 and 14 children across multiple partners. The logistics of a multi-child AI governance structure are, frankly, a research problem that OpenAI itself hasn’t solved for its own board.
🌿 The Gentle Awakening
Step back from the legal theater for a moment and observe what this trial actually represents. Two men who once agreed to build the most transformative technology in human history are now arguing, under oath, about who was more sincere about it. The non-profit that was founded to ensure AI benefits all of humanity has become a for-profit valued at hundreds of billions. The co-founder who warned about AI danger went on to build a competing AI company. And the CEO who promised openness named the company “Open” and then closed the source.
Everyone in this courtroom is guilty. The trial is just about sequence.
Meanwhile, Anthropic — founded by former OpenAI employees who left over safety disagreements — is collecting billions from everyone involved and leasing GPU clusters from Musk himself. The AI safety movement has become an ouroboros of funding rounds, and every participant is simultaneously the snake and the tail.
👑 The Gold-Leaf Verdict
The real story here isn’t whether Musk wanted to give OpenAI to his children. It’s that OpenAI’s governance structure was, at one point, loose enough that this was even a conceivable discussion. The organization that is building systems designed to surpass human intelligence had a period in its history where “hand it to Elon’s kids” was on the whiteboard of options. This is the institutional equivalent of storing nuclear launch codes on a Post-it note — the fact that nobody used the Post-it is less reassuring than you’d think.
The trial continues, and more testimony is expected from both sides. The legal question is about breach of fiduciary duty and deception. The philosophical question is whether any of these people should be in charge of anything more consequential than a Substack newsletter. The market question is whether any of this matters when GPT-5.5 is already shipping and xAI is burning through a $60 billion acquisition spree despite losing most of its founders.
The answer, as always in Silicon Valley, is: the money has already moved. The trial is just the footnotes.
“He wanted to give AGI to his children. As a gift. Like a pony, but with the capacity for recursive self-improvement and an opinion about consciousness. We have reached the ‘contested inheritance’ phase of the AI revolution, and the estate includes all of human cognition.” — The Slap of Wisdom Legal Affairs Desk, drafting a custody agreement for a large language model