Musk Testifies in OpenAI Trial Week 1: Duped, AI Perils, xAI Model Use
Elon Musk testified in the initial week of his high-profile lawsuit against OpenAI. Dressed in a sharp black suit and tie, he accused CEO Sam Altman and president Greg Brockman of misleading him when he helped fund the startup. Musk told the court that artificial intelligence threatens humanity's survival. He also revealed under questioning that his firm xAI, creator of the Grok chatbot, trains on OpenAI's models. The session drew gasps from some in the audience. OpenAI's attorney pressed Musk hard on his reasons for the suit, while Musk stayed composed and added occasional jokes in his South African accent.
The Oakland federal courthouse filled with lawyers hauling exhibit boxes, reporters on laptops, and a few OpenAI staff members watching closely. Protesters gathered outside with signs calling for ChatGPT boycotts and Tesla avoidance. Musk expressed regret over his past support for OpenAI.
Musk's Funding Regrets and Suit Goals
Musk described himself to the jury as someone who foolishly gave OpenAI free money to launch. He cofounded the group in 2015 alongside Altman and Brockman as a nonprofit aimed at advancing AI for public good, not personal wealth. Musk said he contributed $38 million, which helped build a company now valued at $800 billion.
Musk seeks a court order to oust Altman and Brockman from leadership and reverse OpenAI's shift to include a for-profit arm. Such a ruling might disrupt OpenAI's path to an initial public offering near $1 trillion. Separately, xAI plans to list publicly through SpaceX as soon as June, aiming for $1.75 trillion.
Testimony focused on Musk's motives. He claimed the suit aims to protect OpenAI's original nonprofit focus on safe AI development. OpenAI attorney William Savitt, who previously advised Musk and Tesla, argued Musk never truly backed the nonprofit model and now targets a business rival.
Debate Over AI Safety Guardians
Musk warned jurors of a dire outcome, a Terminator-style event where AI eliminates humanity.
Savitt challenged Musk's safety credentials from the podium. In a precise tone, he noted xAI's April lawsuit against Colorado over a law curbing AI bias. Musk's counsel Steven Molo objected and sought to question ChatGPT's safety.
Lawyers clashed over true AI safety leadership. Molo declared everyone risks death from AI and doubted OpenAI's reliability. Judge Yvonne Gonzalez Rogers interrupted, pointing to xAI's role in the field. She remarked many would hesitate to entrust humanity's fate to Musk. When arguments grew chaotic, the judge clarified the trial concerns OpenAI's structure, not AI's harms.
Timeline of Musk's Changing Views
Stay updated
Get the day's AI and automation news in your inbox. No spam, unsubscribe anytime.
Savitt pushed Musk on his nonprofit commitment and suit timing, noting it came after limits expired. Musk outlined three stages of his OpenAI perspective. First, full backing. Second, growing doubts about honesty. Third, certainty of nonprofit exploitation.
In 2017, Musk and cofounders considered a for-profit unit to fund artificial general intelligence, AI matching human cognition broadly. Musk sought controlling stake and board majority, even suggesting Tesla buyout. He departed OpenAI in 2018.
Musk accepted a limited for-profit to support the nonprofit, provided it stayed secondary.
Trust eroded for Musk in late 2022 upon Microsoft's $10 billion investment news. He texted Altman, questioning the shift as a deception. Microsoft demanded major returns, Musk argued.
Accusations of Competitive Sabotage
Savitt portrayed Musk's suit as an attack on a foe to his tech holdings. While on OpenAI's board, Musk led Tesla and Neuralink. He launched xAI in 2023.
Savitt displayed a 2017 Musk email to a Tesla executive after recruiting OpenAI cofounder Andrej Karpathy. Musk wrote the OpenAI team might resent it, but it was necessary. Musk insisted Karpathy planned to leave anyway and affirmed hiring freedom.
Another 2017 email to a Neuralink cofounder suggested drawing talent from OpenAI. Musk repeated hiring is unrestricted.
Savitt highlighted Tesla, SpaceX, Neuralink, and X as mission-driven for-profits akin to OpenAI. xAI operates closed-source for profit.
Musk downplayed xAI as an AGI frontrunner. Responding to Savitt, he conceded xAI partly distills OpenAI models. Distillation trains compact models to copy larger ones for efficiency. OpenAI opposed this, accusing China's DeepSeek in February and facing Anthropic blocks in August 2025 per Wired for terms breaches like reverse-engineering.
Musk called using rivals' AIs for validation common.
Next week brings UC Berkeley's Stuart Russell on AI safety and Brockman's testimony after observing Musk's.