Back to Blog
April 30, 2026 • Tim Fernholz

Musk Confirms xAI Trained Grok Using OpenAI Model Distillation

Musk Confirms xAI Trained Grok Using OpenAI Model Distillation Musk Confirms xAI Trained Grok Using OpenAI Model Distillation Musk Confirms xAI Trained Grok Using OpenAI Model Distillation

Elon Musk testifies that xAI trained Grok on OpenAI models

Key Revelation in OpenAI Lawsuit

During testimony in a California federal court on Thursday, Elon Musk confirmed that his AI company xAI used "distillation" techniques to train Grok by learning from OpenAI's models. When asked directly if xAI had distilled OpenAI models, Musk asserted it was a general practice among AI companies and answered "Partly" when pressed for a yes-or-no response.

What is Distillation?

Distillation is a controversial technique where companies:

  • Systematically query publicly accessible AI chatbots and APIs
  • Use the responses to understand the inner workings of competitor models
  • Train their own models to replicate capabilities at lower cost
  • Effectively copy advanced models without the massive compute investment

Industry Context

The Distillation Problem

  • OpenAI and Anthropic have been actively combating distillation attempts
  • Primary concern has focused on Chinese firms using distillation to create open-weight models nearly as capable as U.S. offerings
  • Tech workers have long assumed American labs use these techniques on each other
  • Musk's admission is the first public confirmation from a major player

Frontier Lab Response

  • OpenAI, Anthropic, and Google launched initiative through Frontier Model Forum
  • Sharing information on combating distillation attempts from China
  • Working to prevent suspicious mass queries
  • Implementing detection systems for systematic model probing

Legal and Competitive Implications

Terms of Service vs. Copyright

  • Distillation not explicitly illegal, but may violate terms of service
  • Creates irony given AI companies' own controversial data practices
  • Undermines competitive advantage from compute infrastructure investments
  • Threatens business models of frontier labs

The OpenAI Lawsuit

Musk is currently suing OpenAI, CEO Sam Altman, and Greg Brockman, alleging they:

  • Breached the original nonprofit mission
  • Improperly shifted to for-profit structure
  • Trial began this week with Musk's testimony

Musk's AI Rankings

Later in testimony, Musk ranked leading AI providers:

  1. Anthropic (top spot)
  2. OpenAI
  3. Google
  4. Chinese open source models
  5. xAI (characterized as "much smaller" with just a few hundred employees)

Industry Impact

This admission reveals:

  • Reality of competitive AI development: Even well-funded startups use shortcuts
  • Fragility of compute moats: Distillation can replicate years of infrastructure investment
  • Hypocrisy in data ethics: Companies that bent copyright rules now fight model copying
  • Arms race dynamics: Frontier labs racing to prevent knowledge transfer while competitors race to extract it