OpenAI Accuses Chinese Rival DeepSeek of Inappropriately Using Its Data
The move follows accusations against OpenAI itself of improperly accessing content it didn't own rights to in building ChatGPT.
OpenAI has raised concerns that DeepSeek, a new Chinese competitor, may have "inappropriately" utilized its data to develop its own artificial intelligence chatbot. DeepSeek's release of an unexpectedly effective and low-cost Large Language Model (LLM) on Monday sent shockwaves through the tech industry, causing shares of Nvidia, the leading U.S. chip manufacturer, to drop significantly.
An OpenAI spokesperson confirmed the company’s suspicions, stating in an email to NBC News that Chinese companies are attempting to catch up to U.S. AI firms by using a process called "distillation." In this method, a new LLM is trained using data generated by another LLM, which OpenAI believes DeepSeek may have used.
“We are aware of and reviewing indications that DeepSeek may have inappropriately distilled our models,” the spokesperson said, noting that the company was not alleging a security breach. Distillation is often prohibited under LLM terms of service, although it is a common practice within the industry.
DeepSeek did not respond to requests for comment.
The situation is complicated by the fact that OpenAI itself has faced allegations of building ChatGPT by accessing content without permission. The company is currently facing multiple lawsuits, including one from The New York Times, which claims that OpenAI used millions of its articles to train its model without consent.
Despite the accusations, OpenAI CEO Sam Altman publicly praised DeepSeek’s work on Monday, calling its model “impressive,” especially considering its affordability. However, OpenAI’s spokesperson hinted that the U.S. government may become involved in the dispute.
“We take aggressive, proactive countermeasures to protect our technology and will continue working closely with the U.S. government to protect the most capable models being built here,” the spokesperson said.
On Tuesday, David Sacks, an AI advisor to President Donald Trump, suggested that DeepSeek had likely engaged in distillation, though he did not comment on potential White House action. “I think over the next few months, our leading AI companies will take steps to prevent distillation,” Sacks told Fox News. “That would definitely slow down these copycat models.”
The White House has not yet responded to inquiries regarding the matter.