Oct 28 (Reuters) – Chinese AI developers DeepSeek and Moonshot AI relied heavily on outputs from OpenAI’s powerful GPT-6 Astra model to train their latest models, The Information reported on Wednesday, citing people with direct knowledge of the matter.
• The practice, known as distillation, has become a flashpoint for leading AI companies as US-China technology competition intensifies, with firms arguing it allows rivals to reproduce advanced capabilities more quickly and cheaply
• Moonshot AI’s Kimi K3.1 Pro model, which was released last week, and DeepSeek’s upgraded V4.1-Flash 1017 model, which was launched on October 17, benefited from distilling Astra, The Information said
• OpenAI, DeepSeek and Moonshot AI did not immediately respond to Reuters requests for comment
• In a September 30 blog post, OpenAI said it had linked a core part of a distillation campaign that began in July to individuals associated with Moonshot AI
• Firms including Anthropic, OpenAI and Google have argued the practice enables rivals to reproduce advanced capabilities more quickly and at lower cost
• Earlier this month, Anthropic said new measures designed to prevent rivals from copying the capabilities of its Claude AI models have reduced data harvested in so-called distillation attacks by about 90% to 99%
(Reporting by Disha Mishra in Bengaluru; Editing by Sherry Jacob-Phillips and Subhranshu Sahu)
