OpenAI president Greg Brockman had an easier second day on the witness stand in Oakland, using questioning from OpenAI's lawyers to reframe personal journal entries that Elon Musk's legal team had presented as a smoking gun during aggressive cross-examination the day before.
Musk's attorney, Steven Molo, had seized on a passage in Brockman's digital diary in which the OpenAI co-founder contemplated a course of action he described as "morally bankrupt."
Under questioning from OpenAI's lawyers, Brockman explained that the passage referred to a plan the company never pursued: removing Musk from OpenAI's board and creating a for-profit subsidiary without him.
Brockman said the co-founders rejected that path and instead spent months trying to find ways to include Musk in any restructuring and offer him equity in the new entity.
Those negotiations collapsed, Brockman testified, when Musk insisted on majority control of the proposed for-profit arm.
When the other co-founders proposed equal equity stakes, Musk declined and demanded to know when Brockman would be leaving OpenAI, threatening to withhold funding.
Brockman described the moment as so tense that he believed Musk was about to physically attack him, testifying that Musk stood up, stormed around the room and grabbed a painting before leaving.
Musk's legal team has argued that his proposed control would have been temporary and would have diluted as new board members and investors joined, but Brockman's account portrayed the breakdown as the moment Musk gave up on OpenAI.
Brockman also used his testimony to question Musk's understanding of the technology at the heart of the case.
He described an episode in which researcher Alec Radford demonstrated an early AI chatbot, a predecessor to the system that would eventually become ChatGPT.
Musk dismissed the demo as "stupid" and told Radford that children on the internet could do a better job, Brockman said, adding that the comment so demoralised Radford that he nearly abandoned AI research entirely and had to be talked back by Brockman and co-founder Ilya Sutskever.
Brockman contrasted that reaction with Sutskever's natural intuition for the technology, recounting a visit to Tesla where Sutskever quickly identified how AI could solve a specific engineering problem, impressing the researchers present.
The trial, which is being heard by a nine-person jury in Oakland federal court, is expected to continue through late May.
Musk is seeking up to $134 billion in damages from OpenAI, Brockman and chief executive Sam Altman, alleging they breached their duties to the nonprofit and enriched themselves by converting OpenAI into a for-profit venture now valued at $852 billion.
Related reading
- Musk drops 24 of 26 claims against OpenAI three days before trial, gambling everything on two
- AI race heats up as GPT 5.5 arrives: OpenAI narrows the gap on efficiency and cost with rivals
- AI can write code and prove theorems. It still can't book me a restaurant
The next witness is expected to be Shivon Zilis, a venture capitalist who served on OpenAI's board until 2023 and is the mother of four of Musk's children.
Her testimony is likely to focus on whether she honoured her fiduciary duties to OpenAI or acted as a conduit for information back to Musk after he left the board in 2018.
