Microsoft is currently conducting an investigation into DeepSeek, a Chinese artificial intelligence startup, over allegations that it may have improperly utilised OpenAI’s application programming interface (API) to extract significant amounts of data.
The inquiry follows claims made by David Sacks, a prominent figure in the AI sector, who asserted that DeepSeek leveraged OpenAI’s models to improve its own AI capabilities.
Reports indicate that Microsoft’s security researchers identified suspicious activities linked to DeepSeek in late 2024, suggesting that the company may have engaged in data exfiltration from OpenAI’s API. As OpenAI’s largest shareholder, Microsoft has taken the initiative to alert OpenAI regarding these potential violations of its terms of service.
OpenAI permits public access to its API. However, its terms explicitly prohibit users from employing the output to develop competing AI models or from extracting data programmatically. The main concern revolves around a method called distillation, where one AI model acquires knowledge from another in a mentor-mentee relationship. If DeepSeek has indeed discovered ways to bypass OpenAI’s rate limits and extensively utilise the API, it may face legal consequences.
DeepSeek has recently gained attention for its R1 reasoning model, which was launched in January 2025 and has reportedly outperformed various leading US AI products on several benchmarks. This fast progression has raised alarms among U.S. officials regarding intellectual property rights and competitive practices in the AI industry. The National Security Council is now reviewing the implications of DeepSeek’s applications, which have recently topped the Apple App Store charts.
In light of these developments, the U.S. Navy has also prohibited the use of DeepSeek’s AI technology due to concerns over security and ethical implications.