Microsoft probing whether DeepSeek improperly used OpenAI APIs

Just a few hours after David Sacks claimed DeepSeek used OpenAI’s models to train its own models, Bloomberg Law reports that Microsoft is investigating DeepSeek’s use of OpenAI’s application programming interface (API).

According to security researchers working for Microsoft, the Chinese company behind the R1 reasoning model may have exfiltrated a large amount of data using OpenAI’s API in the fall of 2024. Microsoft, which also happens to be OpenAI’s largest shareholder, notified OpenAI of the suspicious activity.

While anyone can sign up and access OpenAI’s API, the company’s terms of service stipulate that you can’t use the output to train a new AI model.

“You are prohibited from […] using Output to develop models that compete with OpenAI,” the company writes in its terms of use. Additionally, the company says that you can’t “automatically or programmatically [extract] data or Output.”

The core issue seems to come from distillation, a method used by AI model developers to extract knowledge from another model using a teacher-student sort of behavior. It’s going to be interesting to see whether DeepSeek has found innovative ways to circumvent OpenAI’s rate limits and query its API at scale. If that’s the case, there will certainly be legal ramifications.