The Elon Musk lawsuit against OpenAI has taken a dramatic turn. On the fourth day of the trial, Musk admitted under oath that xAI's Grok model distilled ChatGPT, stating "partially" when questioned by OpenAI's lawyers. Musk later explained that this is something "all AI companies do." As the plaintiff suing OpenAI for "betraying its nonprofit mission," Musk's move has been criticized as a "massive double standard." The trial also revealed Musk's ranking of AI companies: Anthropic first, OpenAI second, Google third, open-source models fourth, and xAI the lowest. Judge Yvonne Gonzalez Rogers refuted Musk's claims about the risks of AI, pointing out that he himself is also building an AI company. OpenAI co-founder Greg Brockman's private diary will be key evidence in next week's trial.
Article author and source: Synced
Elon Musk, who sued OpenAI for "betraying its mission," actually admitted that his Grok model distilled ChatGPT.
Edited by: Peach
[New Intelligence Summary] A major blunder! Elon Musk, who sued OpenAI for "betraying its mission," actually admitted that his Grok model distilled ChatGPT.
On the fourth day of the court battle, Musk dropped a bombshell!
On April 30, the fourth day of the Musk v. OpenAI case took place in the Oakland Federal Court, California.
At the scene, OpenAI's chief counsel, William Savitt, stood up and asked a question that silenced the entire courtroom for three seconds:
Does xAI distill OpenAI's models?
Musk initially dodged the question, saying, "All AI companies are doing this." Savitt pressed further, asking, "So the answer is yes?"
Partly so.
These words stunned the entire internet!
Musk admitted that the Grok model distilled ChatGPT.
A person who is suing OpenAI for "betraying its nonprofit mission" has admitted that his AI distilled a competitor's model.
This blatant double standard has shattered all semblance of decency.
Admittedly, the satire in this scene is overflowing.
In his sworn testimony in court, Musk casually uttered a single sentence that completely shattered the illusion—
This is standard practice in the AI industry; all AI companies do it.
Musk admitted that the Grok model distilled ChatGPT.
Distillation, in essence, is about using a competitor's AI as a tutor to produce a student of similar ability at a low cost.
Musk is suing OpenAI, accusing Altman of violating his nonprofit mission by using the $38 million he donated to start a for-profit company valued at $800 billion.
He is the plaintiff; he is the one who accused others of "stealing charity."
But in court for this very case, he admitted to another form of "theft."
Is distillation illegal? Currently, it's a legal gray area: it may not be illegal, but it almost certainly violates OpenAI's terms of service.
Silicon Valley giants are recklessly skirting the edge of copyright law in their training data, scraping massive amounts of internet content, and then confidently claiming "fair use" when faced with lawsuits from copyright holders.
Anthropic claims that OpenAI violated its terms of service by using its API.
Now even Musk is saying: Everyone's doing it.
This answer was so subtle that it left the lawyer speechless for a few seconds. It acknowledged the situation, but not entirely; it was honest, but left room for interpretation.
Anthropic is first, xAI is last?
Another detail from the trial sparked discussion: Musk personally ranked the global AI companies.
When asked about his bold claim last summer that "xAI will surpass all companies except Google," Musk's ranking in court was:
Anthropic is first, OpenAI is second, Google is third, and open-source models are fourth.
Musk
He said, "xAI is very small, about one-tenth the size of OpenAI, with only a few hundred employees."
This is completely different from his usual style on X.
The man who used to shout "Grok is killing like crazy" on social media suddenly became incredibly humble when he took the oath.
In court, Musk will need to describe xAI as small as possible in order to refute the accusation that "you are suing OpenAI to attack a competitor."
OpenAI Social Media
To put it bluntly: in order to win a lawsuit, you must first admit that you are incapable.
Even harsher than Musk's testimony
One of the most memorable moments of the trial came from Judge Yvonne Gonzalez Rogers.
Musk's lawyers wanted expert witnesses to discuss the possibility that AI could lead to "human extinction," but the judge stopped them on the spot.
Musk himself repeatedly mentioned the "Terminator" scenario in his testimony, saying that he wanted to ensure that Tesla's robots "did not become the Terminator."
Movie screenshot
Who would have thought that the judge's response would be textbook perfect?
I noticed that despite these risks, Mr. Musk himself is creating a company in "this space".
I believe there are many people who are unwilling to entrust the future of humanity to Musk.
But that's not important. This is a case about whether a charitable trust has been violated, not a trial about AI security risks.
The subtext of this statement is far sharper than its literal meaning.
It points to a fact that Musk has been trying to avoid: you shout that AI will destroy humanity while simultaneously founding an AI company, distilling competitors' models, and merging it with SpaceX.
Are you saving humanity or taking business away from them?
OpenAI CEO's Diary: A Real Time Bomb
Musk's testimony was merely the prelude to this trial.
Next Monday, OpenAI co-founder Greg Brockman will testify in court.
Brockman was a key witness because his private diary was accepted as evidence by the court. Several passages from it included:
Ultraman Sam
His story will be: we'll be dishonest with him in the end about still wanting to make it profitable, just without him.
It was wrong to steal the nonprofit from him.
To switch to B-corp without him would be a moral bankruptcy, and he's really not stupid.
These diary entries were one of the key reasons the judge decided to put the case to a jury trial.
OpenAI claims these diary entries were taken out of context by Musk's legal team.
But in any case, Brockman will have to face every word he wrote in court next week.
A trial that rewrote the narrative of the AI industry
Looking back at the four days of court proceedings, the amount of information is astonishing.
Musk admitted to distillation, exposing one of Silicon Valley's biggest open secrets. He ranked AI companies, publicly denigrating his own xAI to the lowest level.
Musk admitted that the Grok model distilled ChatGPT.
He was educated by the judge that "you are not a lawyer" and was forbidden from discussing AI doomsday theories.
His financial manager, Birchall, confirmed the details of the $38 million donation but also revealed that the donation may not have come with any restrictions.
But the most dramatic twist is yet to come; Brockman's witness stand is the real climax of this drama.
Next Monday, the main showdown will officially begin.
