Microsoft investigates if DeepSeek-linked Group improperly obtained OpenAI data | Mint

Big Tech firms, Microsoft Corp and OpenAI, are now investigating whether or not data output from OpenAI’s technology was obtained in an unauthorized way by a group linked to the Chinese startup DeepSeek, reported the news agency Bloomberg according to people aware of the development. 

According to the agency report citing the sources, Microsoft security researchers observed that the individuals who are believed to be allegedly linked to DeepSeek exfiltrated a large amount of data using the OpenAI application programming interface (API). 

The report also said that software developers could pay a license fee to use the API to integrate OpenAI’s proprietary artificial intelligence models into their own applications. 

After this discovery, Microsoft, which is a technology partner of OpenAI, notified the company of this activity, according to the sources. This activity can violate OpenAI’s terms of service or could indicate the group acted to get around OpenAI’s restrictions on how much data they could obtain, according to the agency report.

Microsoft refused to comment on this development, while queries sent to OpenAI remained unanswered. The queries sent to DeepSeek and the hedge fund High-Flyer remained unanswered, according to the news report.

LiveMint couldn’t independently verify the report. 

DeepSeek AI

Earlier this month, the Chinese artificial intelligence startup DeepSeek launched a new open-source AI model called R1 which can mimic the way humans reason, amid a market which is dominated by US rivals like Google, Meta and OpenAI.

According to the agency report, the R1 outperformed the US developers’ products on multiple industry benchmarks from math tasks to general knowledge.

This release tanked the US tech stocks linked to artificial intelligence, like that of companies such as Microsoft, Nvidia Corp., Oracle Corp. and Google parent Alphabet Inc. on Monday, January 27.

US President Donald Trump’s advisor on artificial intelligence, David Sacks, on Tuesday, January 28, said that there is “substantial evidence” that the Chinese AI startup leaned on the output from OpenAI’s models to help develop its AI tech, as per the agency report. 

Sacks explained the technique as distillation, where one AI model uses the outputs of another for training purposes to build similar capabilities, as per the report. 

“There’s substantial evidence that what DeepSeek did here is they distilled knowledge out of OpenAI models, and I don’t think OpenAI is very happy about this,” said Sacks, quoted by the news agency. 

OpenAI, in an official statement, said, “We know PRC-based companies — and others — are constantly trying to distill the models of leading US AI companies.” 

“As the leading builder of AI, we engage in countermeasures to protect our IP, including a careful process for which frontier capabilities to include in released models, and believe as we go forward that it is critically important that we are working closely with the US government to best protect the most capable models from efforts by adversaries and competitors to take US technology,” they said cited the news agency. 

In its own research, DeepSeek said it had “distilled” models from its R1 system based on other open-source systems. Unlike OpenAI’s closed systems, some models, such as Meta’s Llama, are open-source and freely available for use, according to the agency report.