Musk testifies xAI used OpenAI models to train Grok chatbot
Rare admission in federal court as lawsuit over OpenAI's nonprofit mission proceeds
En resumen
- Elon Musk testified in federal court that his AI company xAI partly used OpenAI models to train its Grok chatbot through 'distillation,' a practice under growing scrutiny.
- The admission came during Musk's lawsuit against OpenAI, Sam Altman, and Greg Brockman, alleging the company moved away from its original nonprofit mission.
- Musk co-founded OpenAI in 2015 and left in 2018.
Resumen generado por IA
Por qué importa
The lawsuit centers on Musk's claim that OpenAI moved away from its original nonprofit mission focused on developing AI for the benefit of humanity. Distillation refers to training a new AI system by querying an existing model through its public interface and using outputs as learning signals—a practice not explicitly illegal but potentially violating platform terms.
Elon Musk said his artificial intelligence company xAI partly used OpenAI models while training its Grok chatbot, according to a report by TechCrunch. The admission is a rare public acknowledgment by a major AI developer of a practice under growing scrutiny. It comes as Musk's case against OpenAI moves forward in federal court, where the trial began this week and will examine the company's governance and the broader AI landscape.
Musk made the statement Thursday while testifying in a California federal court, where he is suing OpenAI, CEO Sam Altman, and co-founder Greg Brockman. The lawsuit centers on Musk's claim that OpenAI moved away from its original nonprofit mission.
During questioning, Musk was asked whether xAI used distillation techniques on OpenAI models. He reportedly said the answer was "partly," and described the approach as a broader industry practice.
Musk co-founded OpenAI in 2015 with Sam Altman, Greg Brockman, Ilya Sutskever, John Schulman, and Wojciech Zaremba as a nonprofit focused on developing artificial intelligence for the benefit of humanity. Musk left the company in 2018.
Distillation refers to training a new AI system by querying an existing model through its public interface or API and using those outputs as learning signals. In February, Anthropic accused several Chinese AI developers of using fraudulent accounts to extract large volumes of responses from its Claude chatbot to train competing systems. Earlier this month, the White House warned of "industrial-scale" campaigns using proxy accounts and jailbreaks to replicate U.S. AI capabilities.
Musk's testimony indicates that the method is being used by U.S.-based AI companies, not only foreign competitors. The legal boundaries remain unclear. Distillation is not explicitly illegal, but it can raise questions about whether it violates platform rules or terms governing API use.
Launched in July 2023, xAI entered a market that included Google, Microsoft, and OpenAI, companies with larger teams and more established infrastructure. Earlier that year, Musk and other tech figures signed an open letter calling for a six-month pause on developing more advanced AI systems, citing potential risks.
Musk's remarks suggest the company may have used his former company's technology to close the gap.
Qué observar
Perspectiva de IA — posibilidades, no hechos
The federal court case will continue for several months, examining OpenAI's governance structure
Muy probable · En meses
More AI companies may be scrutinized for using distillation techniques on competitor models
Probable · En meses
Preguntas abiertas
- What specific OpenAI models did xAI use for distillation?
- How much of Grok's training data came from distilled OpenAI models?
- Will the court rule on whether distillation violates API terms of service?
- What are the broader implications for the AI industry regarding distillation practices?
