Everything just stolen?

OpenAI investigates espionage in favor of DeepSeek

According to a report, spied data from ChatGPT may have been used in the development of Chinese competitor DeepSeek.

(Bild: APA/AFP/JOEL SAGET)

The ChatGPT developer OpenAI and Microsoft are investigating suspicions of data theft by Chinese competitors. Last fall, security experts from Microsoft observed the outflow of large amounts of data via an interface of the OpenAI software, Bloomberg reported. Those responsible for this were presumably connected to the Chinese start-up DeepSeek.

Its low-cost yet technologically advanced artificial intelligence caused a stir in the industry at the beginning of the week. David Sacks, who coordinates US President Donald Trump's AI policy, described the theft of intellectual property as possible in a TV interview. "There is solid evidence that DeepSeek has distilled information from OpenAI's AI models."

OpenAI made similar comments when asked and emphasized that Chinese and other companies were constantly trying to copy the AI programs of the leading developers from the USA. OpenAI did not mention the name DeepSeek.

In distillation, existing AI models are used to train new or further developed models. Providers have already improved their own models in this way. What is new is that DeepSeek is said to have used information from third-party models to train its own AI.

OpenAI wants to protect intellectual property with the US government
"We are taking countermeasures to protect our intellectual property," the ChatGPT developer continued. "Going forward, it is critical that we work closely with the US government to best protect the most powerful models from efforts by adversaries and competitors to take over US technologies."

Microsoft did not wish to comment on the possible industrial espionage. DeepSeek could not initially be reached for comment.