Key takeaways:

• Microsoft and OpenAI are probing a potential data breach involving China's DeepSeek.
• Evidence suggests DeepSeek may have used OpenAI's API to exfiltrate data unlawfully.
• DeepSeek's recent AI model launch has led to significant market fluctuations, impacting major tech stocks.

Microsoft Corp. and OpenAI are investigating allegations that a group associated with the Chinese AI startup DeepSeek may have illegally accessed data from OpenAI's proprietary technology. Sources suggest that Microsoft's security team detected suspicious activities linked to DeepSeek in the fall, where individuals were reportedly extracting substantial amounts of information from the OpenAI application programming interface (API). This API allows software developers to utilize OpenAI’s advanced AI models in their own applications, typically under strict licensing agreements.

The investigation comes in light of DeepSeek's recent launch of its open-source AI model, R1, which claims to rival or even outperform established models from OpenAI and other US tech giants like Google and Meta, all while incurring significantly lower development costs. Initial enthusiasm for DeepSeek's technology has triggered a sharp decline in AI-related stocks, with a cumulative loss of approximately $1 trillion in market value for companies including Microsoft, Nvidia, and Alphabet Inc.

According to reports, DeepSeek’s rise has drawn the attention of US officials, with OpenAI noting that Chinese firms “are constantly trying to distill the models of leading US AI companies.” This reflects concerns that DeepSeek may have employed a controversial technique known as "distillation," where one AI model leverages the outputs of another to enhance its own capabilities. David Sacks, US President Donald Trump’s advisor on AI, highlighted ongoing concerns regarding intellectual property theft, stating, “There’s substantial evidence that what DeepSeek did here is they distilled knowledge out of OpenAI models.”

The pressure is mounting on OpenAI and Microsoft to safeguard their innovations against such external threats. Statements from both organizations assert a commitment to protecting their intellectual property and collaborating closely with the US government to reinforce defenses against potential breaches. OpenAI emphasized that it has previously blocked attempts from other entities to use its models without authorization, underlying the significance of this issue for the future of AI development.

As DeepSeek's emergence continues to challenge the US's position in the AI landscape, the investigation underscores the intensifying rivalry in the global tech sector, especially between Chinese and US companies. Concerns over national security and ethical implications further complicate the landscape, with some reports indicating that the US Navy has restricted its personnel from using DeepSeek's applications over these apprehensions.

If you wish to read more about this developing story, visit SCMP, India Times, and The Guardian.