OpenAI and Anthropic have recently taken steps to limit third-party use of their models for distillation, as new testimony from Elon Musk indicates that the practice is also used within the AI industry itself.
Court Testimony Reveals Industry Use Of Distillation
During testimony in a California federal court, Musk said that his company, xAI, has partly used distillation techniques involving OpenAI models to train its Grok system. When asked directly if this was the case, Musk responded, “Partly,” and described the approach as a broader practice among AI companies.
Distillation involves querying existing AI systems to generate outputs that can then be used to train new models, allowing developers to replicate capabilities without the same level of computational investment.
Legal Dispute Centers On OpenAI’s Structure
The testimony comes as Musk pursues legal action against OpenAI, CEO Sam Altman, and co-founder Greg Brockman. Musk alleges that the organization deviated from its original nonprofit mission by transitioning toward a for-profit structure. The trial began this week and includes Musk’s statements under oath.
Distillation Raises Competitive And Policy Concerns
The practice has drawn attention because it can reduce the advantage held by companies that invest heavily in computing infrastructure. By using outputs from established models, other developers may create systems with similar performance at lower cost.
While distillation is not clearly defined as illegal, it may conflict with terms of service that restrict how APIs and chatbot outputs can be used. Concerns have also focused on international competition, particularly regarding lower-cost models developed outside the United States.
AI Firms Coordinate Efforts To Limit Distillation
OpenAI, Anthropic, and Google have reportedly worked through the Frontier Model Forum to share information on detecting and limiting distillation efforts. These measures include identifying patterns of large-scale or automated queries designed to extract model behavior and restricting such activity.
Musk Ranks AI Providers During Testimony
During the same testimony, Musk was asked about earlier statements regarding xAI’s position in the industry. He described Anthropic as leading among AI providers, followed by OpenAI, Google, and Chinese open-source models. He characterized xAI as a smaller organization with only a few hundred employees.
Featured image credits: Flickr
For more stories like it, click the +Follow button at the top of this page to follow us.