Musk Indicates xAI Partly Used OpenAI Models for Training
Elon Musk suggested under oath that xAI has utilized OpenAI's models to some extent in developing its own AI systems. The comment arose during his testimony on Thursday in a federal court case against OpenAI, the creator of ChatGPT. An attorney for OpenAI pressed Musk on the matter during cross-examination.
Details from the Court Exchange
The discussion centered on a technique called distillation. Here is the key back-and-forth, as recorded by reporters:
OpenAI lawyer William Savitt asked: Do you know what distillation is?
Musk replied: It means to use one AI model to train another AI model.
Savitt followed up: Has xAI done that with OpenAI?
Musk said: Generally all the AI companies [do that].
Savitt responded: So that's a yes.
Musk clarified: Partly.
Distillation involves training a smaller AI model to copy the outputs of a larger one. This approach reduces costs and speeds up operations while keeping most of the original capabilities intact.
Savitt then inquired further: Has OpenAI technology been used in any way to develop xAI?
Musk answered: It is standard practice to use other AIs to validate your AI.
Stay ahead of the AI curve
The most important updates, news, and content — delivered weekly.
No spam. Unsubscribe anytime.
Neither OpenAI nor xAI provided immediate comments when contacted by reporters.
OpenAI's Efforts to Protect Its Models
OpenAI has worked to stop rivals from distilling its models. One target has been the Chinese firm DeepSeek. In a February 2026 memo to a House committee, OpenAI stated it had taken steps to protect and strengthen its models against distillation. The company emphasized creating fair conditions where China cannot advance authoritarian AI by copying and reworking U.S. innovations.
The Trump administration has also acted on this issue. Michael Kratsios, director of the White House office of science and technology policy, noted in an April 2026 memo that the government would share details with U.S. AI companies about foreign distillation efforts. Kratsios posted on X that the U.S. government supports open and equitable AI development in a competitive setting.
Practices Among U.S. AI Labs
U.S. AI companies have long used each other's models for evaluation, progress checks, and safety tests. However, competition has led to restrictions. In August 2025, Anthropic denied OpenAI access to its Claude coding models, claiming a violation had occurred. Later, Anthropic also blocked xAI from its coding models.
OpenAI, founded in 2015 as a nonprofit by Elon Musk, Sam Altman, and others, shifted to a for-profit structure in recent years, prompting Musk's lawsuit. Musk, who left the board in 2018, accuses OpenAI of abandoning its original mission. xAI, launched by Musk in 2023, aims to rival OpenAI with models like Grok, focusing on understanding the universe.
Broader Lawsuit Context
During the multi-day cross-examination, William Savitt probed Musk's past efforts to lead OpenAI and his drive to surpass it. The previous day, Savitt showed emails and texts from 2017. These suggested Musk pressured OpenAI by delaying funding and recruiting its top researchers.
The ongoing legal fight highlights tensions in the AI field. Labs compete fiercely for talent and technology amid rapid advances. Musk's testimony sheds light on common yet contentious practices like model distillation, which help firms benchmark but raise intellectual property concerns.
xAI positions itself as a truth-seeking alternative to what Musk calls overly censored AI from competitors. OpenAI defends its pivot to profitability as necessary to fund massive computing needs for models like GPT-4 and successors.
This court revelation adds fuel to the dispute, as both sides vie for dominance in AI development.
Related on Neura Market