Microsoft Probes DeepSeek Over Potential OpenAI API Misuse

Microsoft is investigating whether DeepSeek, a Chinese AI company, improperly used OpenAI’s API to extract large amounts of data, potentially violating OpenAI’s terms of service. The probe follows claims by David Sacks, President Donald Trump’s AI czar, that DeepSeek leveraged OpenAI’s models to train its own AI systems.

According to a report from Bloomberg Law, Microsoft’s security researchers detected suspicious activity from DeepSeek in late 2024, indicating possible exfiltration of OpenAI data. As OpenAI’s largest shareholder, Microsoft flagged the issue directly to the AI company for further examination.

Did DeepSeek Violate OpenAI’s Terms?

While OpenAI allows public access to its API, its terms of service explicitly prohibit:

• Using API-generated output to develop competing AI models

• Automatically or programmatically extracting data from the API

The concern revolves around distillation, a machine learning technique where a smaller AI model learns from a larger one, often using a teacher-student approach. If DeepSeek managed to bypass OpenAI’s rate limits and systematically query the API at scale, it could face legal consequences.

U.S. Officials and OpenAI Respond

Speaking to Fox News, David Sacks claimed there is “substantial evidence” that DeepSeek engaged in distillation using OpenAI’s models. He suggested that U.S. AI companies will take steps in the coming months to prevent such techniques, which could hinder the development of competing AI models.

Despite the allegations, Sacks asserted that the U.S. remains ahead in the AI race. He compared DeepSeek’s R1 model to OpenAI’s o1 model, which launched four months earlier, noting that while DeepSeek is rapidly closing the gap, the U.S. still leads the field.

OpenAI echoed these concerns, stating that China-based companies are “constantly trying to distill” U.S. AI models. The company emphasized its commitment to countermeasures that protect its intellectual property and affirmed its close cooperation with the U.S. government to safeguard advanced AI technologies.

What This Means

If Microsoft confirms that DeepSeek violated OpenAI’s terms, legal action or trade restrictions could follow. However, enforcement would be challenging since DeepSeek operates in China, where U.S. intellectual property laws have little reach. The most immediate consequences may include cutting DeepSeek’s access to OpenAI’s API and strengthening AI security measures industry-wide. This case could further escalate U.S.-China tensions in the AI sector and lead to tighter regulations on how AI models and APIs are protected.

Editor’s Note: This article was created by Alicia Shapiro, CMO of AiNews.com, with writing, image, and idea-generation support from ChatGPT, an AI assistant. However, the final perspective and editorial choices are solely Alicia Shapiro’s. Special thanks to ChatGPT for assistance with research and editorial support in crafting this article.