OpenAI investigates DeepSeek over potential use of ChatGPT data in AI training

iNDICA NEWS BUREAU-

As Chinese artificial intelligence (AI) company DeepSeek continues to make waves in the technology industry amidst the U.S.-China trade tensions, OpenAI has raised concerns that its ChatGPT data may have been used to train DeepSeek’s cost-effective AI models.

OpenAI, led by Sam Altman, has discovered indications that DeepSeek employed a technique known as “distillation,” which is commonly used to train AI models by extracting data from large language models (LLMs). Both OpenAI and Microsoft are now investigating whether DeepSeek utilized their APIs to train its own models.

OpenAI reportedly invested $100 million to train its GPT-4 model. David Sacks, former AI czar under U.S. President Donald Trump, suggested that intellectual property (IP) theft could be involved in DeepSeek’s actions, noting that substantial evidence points to the company distilling knowledge from OpenAI’s models.

OpenAI confirmed that companies based in China and other regions regularly attempt to distill models from leading US AI firms. Meanwhile, the European consumer group coalition, Euroconsumers, has filed a complaint with the Italian Data Protection Authority (DPA) over DeepSeek’s handling of personal data in relation to the General Data Protection Regulation (GDPR).

The Italian DPA has expressed concerns that the data of millions of Italians may be 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 by AI enthusiast Liang Wenfeng in 2015. DeepSeek’s Android app, which serves as an alternative to ChatGPT and is powered by the company’s V3 model, has quickly risen to the top spot on the Google Play Store.

(Photo courtesy: IANS)