The Musk v. Altman trial is generating a steady stream of damaging revelations, and Shivon Zilis keeps appearing at the center of them.
Messages presented at trial show Zilis - a Neuralink executive and mother of four of Musk's children - acting as a conduit between Musk and OpenAI's leadership during a period when the relationship between the two was deteriorating. She appears to have relayed information about internal OpenAI discussions back to Musk while he maintained an outside position on the company.
The dynamic raises questions about the integrity of OpenAI's governance during its formative years. Musk was a co-founder and early funder. If someone close to him was feeding him information about internal strategy and personnel debates, that complicates any narrative that his 2018 departure from the board was a clean break.
For anyone who follows ChatGPT and OpenAI closely, the trial is building a picture of how tangled the early days really were - personal relationships entangled with billion-dollar decisions, informal channels running parallel to official governance. OpenAI was never a normal startup. It was positioned as a safety-first, mission-driven nonprofit, which makes the insider access look considerably worse than it might at a typical tech company.
The lawsuit itself is Musk's attempt to block OpenAI from converting to a for-profit structure. The Zilis thread doesn't directly bear on that legal question, but it hands Musk's legal team a useful narrative: that he had legitimate reasons to feel misled by people he believed were aligned with his vision for the organization.
The court will determine whether the messages show anything actually improper, or just informal communication between people who knew each other personally. But the optics for OpenAI are bad either way - another week of testimony that makes the company's early governance look more like a family drama than a nonprofit board.