OpenAI is investigating a potential data theft incident involving Chinese AI start-up DeepSeek.
Reports indicate that DeepSeek may have illegally accessed data from OpenAI’s ChatGPT.
OpenAI’s largest investor Microsoft is also part of the investigation.
Security experts at Microsoft noticed last fall that individuals linked to DeepSeek were extracting large volumes of data through OpenAI’s application programming interface (API) without authorization, Bloomberg reported.
The unauthorized access could breach OpenAI’s terms of service or suggest that DeepSeek bypassed restrictions on data usage.
OpenAI’s API is used by developers and businesses who integrate its AI models into their applications by purchasing licenses.
The misuse of the API raises serious concerns about data security and intellectual property.
DeepSeek made headlines on Monday when its free AI assistant surpassed OpenAI’s ChatGPT on Apple’s App Store in the United States, causing a drop in tech stocks.
David Sacks, the White House’s AI and crypto advisor, told Fox News that it is “possible” DeepSeek stole intellectual property from the United States.
Sacks said that DeepSeek may have “distilled the knowledge out of OpenAI’s models,” though he did not provide further details.
The Financial Times reported that OpenAI found signs of “distillation,” a process where a complex AI model transfers its knowledge to a simpler one.
This method allows the smaller model to perform similarly with less computing power.
OpenAI suspects DeepSeek of using this technique without permission, which is common but potentially violates their service terms.
An OpenAI spokesperson told Bloomberg that companies, especially those based in China, often try to replicate models from leading American AI firms.
OpenAI said, “We engage in counter-measures to protect our IP and believe it is crucial to work with the US government to safeguard our technology from adversaries and competitors.”
As the investigation continues, both OpenAI and Microsoft aim to secure their data and uphold the integrity of their AI technologies.