OpenAI claims DeepSeek was allegedly trained using its AI models

OpenAI

OpenAI, the artificial intelligence research organisation behind ChatGPT, has announced that it has found evidence suggesting that its models were used to train DeepSeek, the Chinese AI startup. According to a report by the Financial Times, OpenAI discovered that DeepSeek may have employed a technique known as "distillation" to train its AI models. Distillation involves using outputs from larger, more advanced AI models to train smaller, more efficient ones. This method is commonly used in AI development.

As per the report, in late 2024, security researchers at Microsoft discovered a significant amount of data had been leaked from OpenAI developer accounts. Microsoft suspects these accounts are linked to DeepSeek.

In a statement to Bloomberg, OpenAI said that it is aware Chinese companies are constantly trying to distill the models of leading US AI companies. “As the leading builder of AI, we engage in countermeasures to protect our IP, including a careful process for which frontier capabilities to include in released models, and believe as we go forward that it is critically important that we are working closely with the US government to best protect the most capable models from efforts by adversaries and competitors to take US technology.”

US president Donald Trump’s administration has also had their say on the matter. David Sacks, the AI czar for the Trump administration, said that it is possible an IP theft had taken place.“There’s substantial evidence that what DeepSeek did here is they distilled knowledge out of OpenAI models and I don’t think OpenAI is very happy about this,” Sacks told Fox News.

DeepSeek recently gained attention for its DeepSeek-R1 reasoning model, which has achieved capabilities comparable to those of OpenAI's models at a fraction of the cost. The company claims to have spent only $5.6 million on development, compared to the hundreds of millions invested by OpenAI and other leading AI firms.