OpenAI probing if China’s DeepSeek used its models to train AI

As Chinese AI company DeepSeek gains traction in the tech world amid ongoing US-China trade tensions, OpenAI suspects that DeepSeek may have utilized ChatGPT data to develop its own cost-effective AI models.

OpenAI, led by Sam Altman, informed the Financial Times that it has found evidence suggesting DeepSeek employed a technique known as “distillation,” a common method used to extract knowledge from large language models (LLMs) to train new AI models.

Both OpenAI and Microsoft are now investigating whether DeepSeek leveraged their APIs to train its own AI systems.

OpenAI reportedly invested $100 million in training its GPT-4 model.

David Sacks, artificial intelligence advisor to former US President Donald Trump, stated that potential intellectual property theft may have occurred in this case.

“There is substantial evidence that DeepSeek extracted knowledge from OpenAI’s models, and I doubt OpenAI is pleased about it,” Sacks told Fox News.

In a statement, OpenAI noted, “We are aware that PRC (China)-based companies—and others—are constantly attempting to distill models developed by leading US AI firms.”

Meanwhile, Euroconsumers, a European coalition of consumer groups, has filed a complaint with Italy’s Data Protection Authority regarding DeepSeek’s handling of personal data and its compliance with GDPR regulations.

The Italian regulator warned that “the data of millions of Italians is at risk” and has given DeepSeek 20 days to respond.

DeepSeek is backed by High-Flyer Capital Management, a Chinese quantitative hedge fund co-founded in 2015 by AI enthusiast Liang Wenfeng.

At the same time, DeepSeek’s Android app has soared to the top of the Google Play Store, positioning itself as a ChatGPT alternative powered by the Chinese firm’s V3 model.