OpenAI accuses DeepSeek of stealing its proprietary AI models to train its open-source AI using ‘distillation’ technique

OpenAI says it has found proof that DeepSeek illegally used its proprietary models to train an open-source competitor, according to a report from the Financial Times.

The maker of ChatGPT claims DeepSeek used a method called “distillation,” which allows a smaller model to replicate the performance of a larger one with fewer computing resources. OpenAI argues that this constitutes unauthorized use of its technology, raising concerns over potential intellectual property theft.

OpenAI Says it Has Evidence China’s DeepSeek Used its Model to Train Its Open Source AI

The San Francisco-based AI company told the Financial Times that it had gathered evidence suggesting DeepSeek leveraged distillation to improve its AI models. This method, commonly used in machine learning, enables developers to refine smaller models by using outputs from more advanced systems.

“OpenAI says it has found evidence that Chinese artificial intelligence start-up DeepSeek used the US company’s proprietary models to train its own open-source competitor, as concerns grow over a potential breach of intellectual property,” Financial Times reported.

The Financial Times continued:

“The San Francisco-based [ChatGPT] maker told the Financial Times it had seen some evidence of distillation which it suspects to be from DeepSeek. The technique is used by developers to obtain better performance on smaller models by using outputs from larger, more capable ones, allowing them to achieve similar results on specific tasks at a much lower cost.”

In response, OpenAI said it has ramped up collaboration with the U.S. government to protect American AI technology from foreign threats, particularly from China. The company stressed the importance of securing intellectual property and preserving the advanced capabilities built into its models.

DeepSeek’s AI launch has already shaken the market, contributing to a massive $700 billion wipeout in tech stocks. The company’s cost-efficient model poses a direct challenge to U.S. firms that have dominated the AI space.

OpenAI CEO Sam Altman acknowledged DeepSeek’s model as a strong competitor and pointed to the need for greater computing power to maintain an edge. Former President Donald Trump weighed in on the situation, stressing the importance of keeping the U.S. at the forefront of technology.

Despite the shake-up, some analysts see an upside. They believe DeepSeek’s efficiency gains could push AI adoption even further.

According to multiple reports, DeepSeek V3 outperformed leading models like Llama 3.1 and GPT-4o on key benchmarks, including competitive coding challenges on Codeforces. The project was completed on a budget of just $5.5 million—a stark contrast to the hundreds of millions spent by its rivals. This breakthrough challenges the notion that cutting-edge AI development requires an enormous financial investment.

Meanwhile, the big question remains: If OpenAI’s model is closed-source, how did DeepSeek get access to it? The other question is: If DeepSeek stole OpenAI’s proprietary model, as the company claims, why isn’t OpenAI performing better than DeepSeek?