OpenAI and Anthropic are currently taking action against third-party organisations attempting to use their chatbots and APIs for creating new AI models, a process known as “distillation.” This practice has gained attention, particularly regarding Chinese companies, which are leveraging distillation to develop competitive open-weight models at significantly lower costs compared to those from U.S. firms. Many industry insiders have suspected that American tech labs are also engaging in distillation to keep up with rivals.
This suspicion was confirmed during a federal court hearing in California, where Elon Musk testified about his company, xAI. He acknowledged that xAI has employed distillation techniques on OpenAI’s models while training its own product, Grok, suggesting that this is a common practice across AI companies.
Musk is currently involved in legal action against OpenAI’s CEO Sam Altman and co-founder Greg Brockman, claiming they have strayed from OpenAI’s original nonprofit mission by adopting a profit-driven structure. This trial features Musk’s testimony and highlights the competitive tension in the AI sector.
The implications of Musk’s admission are significant, as distillation poses a risk to leading AI companies by enabling others to produce nearly equivalent models without the heavy investments in computational power that these giants have made. Ironically, while frontier labs like OpenAI are exploring various means, including contentious copyright practices, to amass sufficient data for training their models, they are also working to safeguard their technologies from distillation.
In response to the growing concern over distillation, which often involves extensive queries to decipher the internal workings of models, OpenAI, Anthropic, and Google have united through the Frontier Model Forum to combat efforts emanating from China. They aim to implement measures that restrict suspicious bulk querying by users.
As of the latest reports, OpenAI has not commented on Musk’s statements. Additionally, Musk discussed his expectations for xAI’s competitive positioning in the AI landscape, stating that while he had previously claimed xAI would surpass all competitors except Google, he now ranks leading AI entities as follows: Anthropic at the top, followed by OpenAI, Google, and then Chinese open-source models. Notably, he described xAI as a relatively smaller company, with only a few hundred employees.
Through these developments, the scrutiny on distillation practices in the AI domain intensifies, revealing both competitive strategies and potential legal challenges that may shape the future of AI technology.
Fanpage: TechArena.au
Watch more about AI – Artificial Intelligence
