ChatGPT Maker Suspects China’s Dirt Cheap DeepSeek AI Models Were Built Using OpenAI Data — and the Irony Is Not Lost on the Internet

OpenAI has voiced concerns that China's DeepSeek AI models, known for their low cost, may have been developed using data from OpenAI. This week, Donald Trump called DeepSeek a wake-up call for the U.S. tech industry, following a significant drop in Nvidia's market value—nearly $600 billion—triggered by DeepSeek's emergence. The launch of DeepSeek caused a sharp decline in the stock prices of major AI-focused companies. Nvidia, a leading supplier of GPUs crucial for AI model operation, experienced the steepest fall, with a 16.86% drop—the largest single-day loss in Wall Street history. Microsoft, Meta Platforms, Alphabet, and Dell Technologies also saw substantial losses.

DeepSeek claims its R1 model is a significantly cheaper alternative to Western AI models like ChatGPT, built upon the open-source DeepSeek-V3. This model reportedly requires less computing power and was trained for an estimated $6 million—a figure disputed by some. However, DeepSeek's impact has raised questions about the massive investments American tech companies are making in AI, unsettling investors. DeepSeek's popularity surged, becoming a top downloaded free app in the U.S. amidst growing discussions about its performance.

Bloomberg reported that OpenAI and Microsoft are investigating whether DeepSeek utilized OpenAI's API to integrate OpenAI's AI models into its own. OpenAI stated that they are aware of attempts by Chinese and other companies to leverage leading U.S. AI models and are implementing countermeasures to protect their intellectual property. This includes careful selection of capabilities included in released models and close collaboration with the U.S. government.

David Sacks, President Donald Trump's AI czar, suggested that DeepSeek may have used a technique called distillation to extract data from OpenAI models, a violation of OpenAI's terms of service. He anticipates further measures from leading AI companies to prevent similar incidents.

The situation highlights a degree of irony, given OpenAI's own past controversies. OpenAI has previously acknowledged its reliance on copyrighted material for training ChatGPT, stating in a submission to the UK's House of Lords that it would be impossible to create today's leading AI models without using copyrighted materials. This stance is further underscored by the ongoing lawsuits against OpenAI, including one from the New York Times for alleged unlawful use of its work and another from 17 authors accusing "systematic theft on a mass scale." OpenAI maintains that its training practices constitute "fair use." The legal landscape surrounding the use of copyrighted material in AI training remains complex, particularly in light of a 2018 U.S. Copyright Office finding that AI-generated art is not copyrightable.