Musk Admits xAI Used OpenAI Models to Train Grok
Musk Testifies xAI Trained Grok Using OpenAI Model Distillation
Elon Musk admitted under oath on April 30, 2026, that xAI used distillation techniques on OpenAI's models to help develop Grok — a disclosure that landed in the middle of his own lawsuit against OpenAI, CEO Sam Altman, and co-founder Greg Brockman. The trial, which began the week of April 28, 2026, in a California federal court in Oakland, centers on Musk's claim that OpenAI breached its original nonprofit mission by transitioning to a for-profit structure. The courtroom admission adds a significant layer of complexity to a case Musk has framed as a defense of AI safety and nonprofit integrity.
When OpenAI's attorney William Savitt asked Musk whether xAI had used OpenAI's models to help develop or test its own models, Musk confirmed it. Asked directly whether that answer meant "yes," Musk replied, "Partly." He went on to assert that using distillation techniques on other companies' models was a general practice across the AI industry — a defense that industry observers will scrutinize closely given the current climate around unauthorized model distillation.
What Is Distillation — and Why Does It Matter?
Distillation, in the context of artificial intelligence, typically involves systematically querying a larger "teacher" model to understand its inner workings and use that knowledge to train a smaller, cheaper "student" model. When applied to a company's own models, the technique is widely used and legally uncontroversial. When applied to a competitor's models without authorization, it is considered a violation of terms of service — and, increasingly, a national security concern.
According to TechCrunch, it is not clear that distillation is explicitly illegal, but it may violate the terms of service companies set for the use of their products. To combat unauthorized distillation, frontier labs are working to prevent users from making suspicious mass queries against their systems.
Musk's admission is particularly striking because xAI, the company he founded in 2023, is a direct commercial competitor to OpenAI. Tesla, which has invested $2 billion into xAI, has integrated a version of its Grok AI chatbot into Tesla vehicles' media and navigation systems, making Grok a consumer-facing product with real commercial stakes.
The Frontier Model Forum and the Distillation Arms Race
Musk's courtroom disclosure arrives as the AI industry is in the midst of a coordinated effort to combat large-scale distillation attacks, particularly from Chinese AI firms. According to TechCrunch, OpenAI, Anthropic, and Google have launched an initiative through the Frontier Model Forum to share information about how to combat distillation attempts from China. The Frontier Model Forum was founded in 2023 by OpenAI, Anthropic, Google, and Microsoft as a self-regulatory body, and has more recently been used to coordinate anti-distillation efforts.
The scale of documented attacks underscores why the industry has mobilized. In February 2026, Anthropic estimated that three Chinese AI firms — DeepSeek, Moonshot AI, and MiniMax — collectively generated over 16 million exchanges with Claude from around 24,000 fraudulently created accounts. Of those three firms, MiniMax alone drove over 13 million exchanges. Anthropic described AI companies as entities that "routinely distill their own models to create smaller, cheaper versions" — a framing that drew a clear line between legitimate internal distillation and the unauthorized external variety attributed to those Chinese firms.
Microsoft, notably, is both a founding member of the Frontier Model Forum and a named defendant in Musk's lawsuit. Musk accuses Microsoft of aiding and abetting OpenAI, Altman, and Brockman's alleged misconduct in the shift to a for-profit structure.
Musk's Testimony: Rankings, Company Size, and a Safety Dispute
Beyond the distillation admission, Musk's April 30 testimony offered a candid self-assessment of where xAI stands in the competitive landscape. According to TechCrunch, Musk ranked the world's leading AI providers during his testimony, placing Anthropic at the top, followed by OpenAI, Google, and Chinese open source models. He characterized xAI as a much smaller company with just a few hundred employees — a striking contrast to the scale of operations at the companies he ranked above it.
OpenAI's lead attorney William Savitt described Musk as a "reluctant witness" in a meeting with reporters after court adjourned on April 30, according to Axios.
Musk's testimony also touched on AI safety, a central theme of the case given his public positioning as a safety-conscious voice in the industry. When pressed on safety-related questions, Musk responded at one point: "Safety card? Why would it be a card?" He also offered a broader comment on the nature of AI-generated content and its influence on users: "Just because you may read something that is racist or sexist doesn't mean you'll become racist or sexist."
Context: Why This Admission Matters for the AI Industry
The irony embedded in Musk's testimony is difficult to overstate. He filed a lawsuit against OpenAI in part on the grounds that the organization deviated from principles it was founded on. At the same time, his own company — xAI — used distillation techniques on OpenAI's models to develop a competing product. Musk's defense, that this is standard industry practice, is not without basis; distillation from one's own models is indeed routine. But the line between permissible and impermissible distillation is precisely what the industry is now trying to enforce, and Musk's characterization of distillation from competitors' models as broadly normal will likely be contested.
The trial itself began the week of April 28, 2026, with Musk taking the stand in Oakland on April 28. Microsoft is also named as a defendant, with Musk alleging it aided and abetted the alleged breach of charitable trust by OpenAI's leadership. Musk co-founded OpenAI with Altman in 2015 as a nonprofit and departed in 2018 before founding xAI in 2023.
In February 2026, xAI was acquired by SpaceX, becoming a subsidiary as SpaceX consolidated its AI operations ahead of a planned IPO, according to publicly available records. That structural backdrop adds another dimension to questions about xAI's governance and mission — questions that mirror, in some respects, the ones Musk is raising about OpenAI in court.
What Comes Next
The Musk v. OpenAI trial is ongoing as of April 30, 2026. The distillation admission will likely factor into how both sides frame their arguments going forward: OpenAI's legal team may use it to challenge Musk's standing as a principled critic of AI industry practices, while Musk's defense will likely continue to press the case that distillation is an accepted norm across the industry.
On the broader distillation front, the Frontier Model Forum's coordinated effort to share threat intelligence represents the industry's primary institutional response to unauthorized model copying. Whether that effort produces enforceable standards — or remains a voluntary coordination mechanism — remains to be seen. The legal status of distillation itself is still unsettled; what is clear is that terms of service violations, not criminal statutes, are currently the primary enforcement mechanism available to frontier labs.
For companies and individuals tracking the AI landscape, the Musk v. OpenAI trial is shaping up to be a defining moment — one that surfaces questions about model ownership, competitive practice, and the norms that govern how AI systems are built at the frontier.
For more tech news, visit our news section.
Stay Ahead of the AI Curve with Moccet
Understanding how AI models are built, trained, and deployed isn't just a concern for lawyers and tech executives — it directly shapes the tools that affect your productivity, focus, and daily decision-making. As the AI landscape evolves at speed, staying informed gives you an edge in choosing the platforms and systems you trust. Join the Moccet waitlist to stay ahead of the curve.