OpenAi And Microsoft Concerned Stolen Data To Develop DeepSeek

DeepSeek, a rising AI competitor from China, is making waves in the U.S. market, and not all of them are welcome. Concerns around data privacy and intellectual property have surfaced, especially as DeepSeek claims its AI models were trained with significantly fewer resources than its rivals. According to a recent report from Bloomberg, Microsoft and OpenAI are investigating whether DeepSeek improperly accessed OpenAI’s data.

Allegations of Data Misuse

Microsoft’s security team detected unusual activity in late 2024, where developer accounts possibly linked to DeepSeek were reportedly exfiltrating large amounts of data via OpenAI’s API. While the OpenAI API allows developers to integrate its models into their applications, distilling outputs for independent model development violates OpenAI’s usage policies.

David Sacks, President Donald Trump’s AI czar, shared his concerns during an interview with Fox News:

“There’s substantial evidence that DeepSeek essentially distilled knowledge from OpenAI’s models to cheaply develop its own technology… AIs can ask millions of questions and mimic the reasoning process of a parent model. OpenAI is understandably not happy about this.”

OpenAI echoed similar concerns, releasing a statement to Fox News: