Elon Musk confirmed in federal court that his startup xAI used OpenAI's models to train its Grok artificial intelligence system. The testimony came during an ongoing legal dispute between the two firms. Musk stated xAI employed knowledge distillation techniques to transfer capabilities from OpenAI's models to Grok.
Direct Court Admission
Musk acknowledged under oath that xAI leveraged OpenAI's models during Grok's development phase. The process involved knowledge distillation, where a smaller model learns from a larger model's outputs. He did not specify which OpenAI models were used or the time frame for training.
Distillation Technique Explained
Knowledge distillation is a standard machine learning method where a new model mimics an existing system's behavior. Grok's creators applied this approach using OpenAI's technology as the source. The technique doesn't copy underlying code but replicates responses and patterns.
Legal Case Context
The admission occurred during federal court proceedings initiated by OpenAI. The lawsuit centers on xAI's use of OpenAI's intellectual property. Musk's testimony directly addresses the core allegation in the case without detailing damages or remedies.
Immediate Implications
Grok remains fully operational for X Premium users despite the testimony. xAI has not adjusted the model or training approach since the court appearance. No changes to user access or functionality have been announced.
Both companies will file additional evidence by month's end as the case moves toward discovery phase.
