A super AI giant valued at nearly a trillion dollars is having its dirty laundry aired in a California courtroom. The landmark trial brought by Elon Musk against OpenAI CEO Sam Altman has entered its second week. What began as a serious commercial lawsuit concerning "humanity's fate and AI safety" has devolved into a spectacle of executive mudslinging and personal attacks. The judge has stated bluntly that this trial ultimately boils down to which side the jury chooses to believe. Based on current court records, OpenAI's former core founding team is laying bare their worst sides for the world to see.

**Former CTO Exposes Altman's "Toxic Culture" and Deception** In a moment that has shocked observers, Mira Murati, OpenAI's former technical core who led the releases of GPT-4 and ChatGPT, broke her silence in court. Her sudden departure in 2024 sparked widespread speculation, and in her video deposition, she directly targeted her former boss, Altman. Murati testified that Altman lied to her directly about the safety review process for a new model. He allegedly told her that Jason Kwon, the head of legal, said the new model did not need review by the safety committee. However, when Murati verified this herself, she found their accounts "did not match at all." Murati stated in court that Altman fostered a culture of "chaos and distrust" within the executive team. She described him as skilled at using shrewd management tactics, selectively disclosing key information, providing different versions of facts to different executives, and obfuscating safety review checkpoints—all to ensure final decision-making authority remained firmly in his own hands. Former board members Tasha McCauley and Helen Toner, who were involved in Altman's brief ousting, submitted similar testimonies. They characterized the corporate culture under Altman as a "toxic culture" rife with "chronic lying and deception." McCauley stated that Altman's dishonesty triggered a "crisis event" every few months and that he even spread false rumors to manipulate the board. Former chief scientist Ilya Sutskever reportedly detailed the chaos caused by Altman's various lies in a multi-page email. As previously reported, Anthropic CEO and OpenAI co-founder Dario Amodei wrote in his private notes: "The problem at OpenAI is Sam himself."

**Counterattack Against Musk: An "Ignorant and Volatile" Figure** In response to Musk's aggressive allegations, Altman's camp launched a fierce counterattack, focusing on Musk's professional competence and emotional control. OpenAI's current president, Greg Brockman, testified that during the internal power struggle at OpenAI in 2017, a major concern was Musk's fundamental lack of the technical capability to lead AI development. Brockman did not mince words, testifying: "Look, he understands rockets, he understands electric cars, but he does not understand artificial intelligence, and I believe he still doesn't." He pointed out that Musk was unwilling to invest the time to truly master AI technology. Brockman also vividly described Musk's "volatile temper" to the jury. He recalled a meeting in August 2017 discussing restructuring, during which Musk suddenly lost his temper. "He stood up and started pacing angrily around the table. I genuinely thought he was going to hit me, genuinely thought he would physically attack me." Ultimately, Musk grabbed a recently received painting gift and stormed out, slamming the door. As previously reported, testimony this Wednesday from Shivon Zilis, Musk's former partner and a former OpenAI board member, also corroborated the breakdown in communication among executives. She admitted that Musk and the founding team were "sometimes not good at communicating" and that for difficult issues, they had to specifically choose moments when Musk was "in a good mood and clear-headed" to approach him. Zilis further testified that Musk once wanted to merge OpenAI into Tesla Motors and offered Altman a seat on Tesla's board as an incentive. This testimony touches upon the core allegation by Musk that OpenAI betrayed its promise of a non-profit mission.

**A $100 Million Pressure Point and a "Captured" Mission** Beyond the executive drama, the trial has revealed how OpenAI gradually strayed from its non-profit mission of "benefiting all of humanity" and became subservient to capital. The trial exposed a critical timeline: in 2022, before ChatGPT's release, Microsoft CEO Satya Nadella explicitly told Altman that OpenAI needed to achieve $100 million in revenue to qualify for Microsoft's next $10 billion investment. To secure that $10 billion, Altman accepted the condition and made it a core internal pressure point. Former CTO Murati confirmed that the company began massively expanding its commercial system to hit that number. In the face of commercial interests, AI safety became a complete afterthought. Former safety researcher Rosie Campbell testified that as the company became increasingly product-centric, two teams dedicated to long-term AI safety were ultimately disbanded. Adding to the irony is OpenAI's "Deployment Safety Board (DSB)." This body, meant to oversee the safe release of AI models, was designed in a way that facilitated "collusion": the six-member board comprised three Microsoft and three OpenAI representatives, with four votes needed for approval. Altman was one of the three OpenAI representatives. This meant that "3 Microsoft members + Altman = 4 votes," allowing any model to be forcibly deployed, rendering internal OpenAI safety dissenters powerless. To bypass the process, Altman even lied directly to Murati, claiming GPT-4 Turbo did not require DSB review.

**A Non-Profit Shell in Name Only** Musk's assertion from years ago, that "OpenAI is not open anymore," seems to be confirmed in court. Today, the non-profit organization has become an empty shell with almost no real power or assets, holding only a 26% stake in the for-profit company. All top talent, core technology, and valuable assets have been transferred to the for-profit entity. A surreal scene played out in the courtroom. During his five-hour testimony, OpenAI President Brockman personally admitted to every element of "unjust enrichment" and "breach of fiduciary duty." However, when the plaintiff's attorney asked if he knew why he was being sued, he feigned ignorance before the jury, repeatedly claiming he "didn't know." When the attorney accused him of "violating the charitable mission to develop safe AI in an open-source manner for the benefit of humanity, not for personal profit," Brockman could only offer a weak rebuttal: "We didn't do that." Attorney: Has this lawsuit prevented you from publishing a new charter? Brockman: Honestly, I've never been quite sure what I'm being sued for. Attorney: You're not sure? Have you read the complaint? Brockman: Very carefully. Attorney: They hired some very famous lawyers, a well-known law firm—and they never explained to you why you're being sued? Brockman: I understand the allegations. I've also seen how they've changed over time. But— Attorney: So you do understand? Brockman: I stand by my testimony. Attorney: I thought you said you didn't understand why you were being sued. Brockman: That is indeed what I said. Yes. Attorney: And you stand by that testimony? Brockman: Yes. Subsequently, Musk's attorney directly stated the accusation: Attorney: You are being sued because you breached your fiduciary duty to the charity by failing to follow the mission you just had on your website announcing to the world. Understood? Brockman: Oh, I just disagree with that. Attorney: You're not in a position to disagree. I'm asking you. You said you didn't know why you were being sued, so I'm telling you: you are being sued for violating OPENAI's charitable mission to develop safe AGI on an open-source basis for the benefit of humanity, with no individual profiting. Do you understand that is why you are being sued? Brockman: That is not what we did. This layer-by-layer courtroom debate has revealed the full picture of the power struggles, filled with lies and calculation, within OpenAI. Next week, Altman, at the center of the storm, will take the stand to testify personally. The most expensive and naked power game in tech history is about to reach its climax.