Microsoft Bans Employees From Using DeepSeek, Despite Hosting It On Azure

Microsoft has taken a firm stance against the use of the Chinese AI tool DeepSeek within its organization, officially banning employees from using the technology in workplace settings or on corporate-issued devices. The decision is rooted in concerns over data security and potential risks linked to foreign-developed artificial intelligence platforms, particularly those originating from China.

Brad Smith, president of Microsoft, confirmed the ban and stated that DeepSeek would not be made available on the Microsoft Store. The reasoning behind this is to avoid not only any security vulnerabilities but also to discourage what he referred to as “Chinese propaganda” associated with the tool’s usage. Smith made these remarks during a recent U.S. Senate hearing, where he voiced strong support for U.S. government efforts—especially those backed by the Trump administration—to restrict the use of Chinese-developed AI systems on federal networks and devices.

Despite this restrictive internal policy, Microsoft has made the DeepSeek R1 model accessible through its Azure AI Foundry platform. This move has drawn attention due to its apparent contradiction with the company’s internal ban. However, Microsoft has clarified that DeepSeek is an open-source AI model, which allows it to be hosted on private servers without requiring any communication with Chinese-based infrastructure. This key distinction significantly reduces the security risks often associated with foreign AI systems, such as data leakage or unauthorized access.

When asked about this inconsistency, Smith explained that Microsoft had implemented several modifications to the DeepSeek model before making it available via Azure. According to him, these “cuts” were necessary to eliminate “harmful side effects,” thereby enabling a more secure and controlled version of the AI to be offered to developers and researchers. This approach reflects Microsoft’s attempt to balance access to open-source innovation with the need to maintain strict cybersecurity standards.

In summary, Microsoft’s dual approach to DeepSeek—banning its use internally while still providing a curated version on Azure—demonstrates the complexities of navigating AI development in a global and politically sensitive landscape. The company appears committed to supporting innovation, including tools developed outside the U.S., but only under conditions that ensure data protection and compliance with national security priorities. This case highlights broader tensions between open-source accessibility and geopolitical caution in the evolving world of artificial intelligence.

Filed in Robots. Read more about Microsoft.