OpenAI has alerted U.S. policymakers about concerns that DeepSeek, a Chinese AI company, is mimicking American AI models. An internal memo obtained by Reuters highlights that DeepSeek is allegedly circumventing technical restrictions to extract knowledge from leading U.S. systems, ultimately aiming to enhance its own AI models. OpenAI claims that DeepSeek employees have developed dubious methods for accessing AI models and using scripts for automated knowledge collection, a process known as distillation, which allows less advanced models to learn from more sophisticated ones rapidly.
CEO Sam Altman argues that these practices represent unfair competition, infringing upon the investments of American AI labs. Despite the lack of a response from DeepSeek or its parent company, High-Flyer, there is recognition in tech circles of DeepSeek’s technical capabilities. OpenAI emphasizes the necessity for responsible AI development, stressing that geopolitical tensions are increasingly influencing the global AI race.
Source link
