Musk v. Altman in court: alleged deception, an AI apocalypse warning and OpenAI distillation

The historic confrontation between Elon Musk and OpenAI has moved from Twitter spats to federal court. The first week of hearings produced three sensational moments: an accusation of deception, a warning about an AI apocalypse, and an admission that xAI copies a competitor's models.

Musk on the witness stand

Musk arrived in court in a black business suit and tie and took the witness stand, which is unusual for a plaintiff by American standards. His testimony rested on one key claim: Sam Altman and OpenAI president Greg Brockman deliberately misled him when they brought him in to fund the company at an early stage.

According to Musk, the original agreement was verbal: OpenAI would remain a nonprofit whose sole purpose was to develop safe AI for the benefit of all humanity, not of individual shareholders. Musk put in about $44 million in the early years, counting on exactly that model.

He claims that OpenAI's transformation into a commercial company valued at more than $300 billion directly violates those agreements, and that his money and public support helped Altman build all of it.

Altman denies any part in the alleged deception. His side insists that Musk understood the organization could change, and that a nonprofit structure is not a permanent fixture.

AI will destroy humanity, according to Musk

Alongside the legal arguments, Musk used the stand to repeat one of his signature theses about AI risk. In his words, it was precisely the fear of uncontrolled AI that meant safety should have come first at OpenAI, ahead of commercial success.

• Musk said AI poses a real threat of destroying humanity
• OpenAI was created as a "counterweight" to commercial AI from Google and Meta*
• Microsoft invested more than $13 billion and gained substantial control
• Musk believes his name and money were used to legitimize the project
• He himself launched xAI and Grok, which creates an obvious conflict of interest in the case

Apocalyptic rhetoric is nothing new for Musk: he signed open letters calling for a pause in AI development and has talked about superintelligence risks for years. Critics, however, point out that he is actively building commercial AI himself and competes for the same resources and talent as OpenAI.

The surprise: xAI distills OpenAI models

The most unexpected turn of the week was an admission during cross-examination: xAI uses distillation of OpenAI models in developing Grok. Distillation means training a new language model on the outputs of a more powerful system, effectively adopting its behavior. It is a legal practice, but it creates obvious awkwardness: Musk is suing OpenAI while using its models as a teacher for his own products.

"xAI distills

OpenAI models": the admission was made in court and immediately became the top quote of the first week.

The defendants' lawyers are already using it in their defense. The admission complicates the "Musk versus bad OpenAI" narrative: the plaintiff, it turns out, willingly uses the very thing he attacks in court.

What it means

Musk v. OpenAI is not just a conflict between two billionaires. It is a precedent for the whole industry: can courts force AI companies to stick to their original mission, what do verbal agreements mean in a world of big money and big models, and who ultimately bears responsibility when a lab idea turns into a trillion-dollar business.

*Meta is designated an extremist organization and banned in Russia.