Microsoft employees are banned from using DeepSeek app, president says

: Microsoft's Brad Smith announced in a Senate hearing that their employees are prohibited from using the DeepSeek app, as Microsoft is worried about the security risks related to data potentially being stored in China, as well as propaganda concerns from China's government influencing the app's outputs. Despite these concerns and not listing DeepSeek in its app store, Microsoft hosted DeepSeek's R1 model on Azure cloud service earlier this year after evaluating its safety. The DeepSeek app allows data to be stored on Chinese servers as per its privacy policy, and Microsoft has managed to alter DeepSeek's AI model to curtail harmful effects. Interestingly, while DeepSeek is banned, Perplexity—a competing chat app—remains accessible in the Windows store despite the competitive landscape with Microsoft's own Copilot app.

Microsoft's vice chairman and president, Brad Smith, has publicly declared that Microsoft employees are forbidden from using the DeepSeek application, highlighting concerns over data security and the potential spread of Chinese propaganda. This decision was revealed during a Senate hearing on strengthening U.S. capabilities in computing and innovation. Smith noted that DeepSeek's privacy policy states user data is stored on Chinese servers, which are subject to Chinese laws that require cooperation with the country's intelligence agencies.

Microsoft has not added DeepSeek to its app store, further emphasizing their caution regarding the application. This move aligns with decisions made by various organizations and countries that have imposed their own restrictions on DeepSeek. Despite DeepSeek's open-source nature, which allows for its model to be downloaded and employed without routing data back to Chinese servers, concerns remain about the app potentially generating propaganda or insecure code.

Interestingly, Microsoft has hosted DeepSeek's R1 model on its Azure cloud service even after the model gained significant attention earlier in the year. Before offering it on Azure, DeepSeek underwent rigorous red teaming and safety evaluations, ensuring that it aligns with Microsoft's safety standards. Smith mentioned that Microsoft has accessed DeepSeek’s AI model and implemented changes to mitigate harmful effects, though specifics of these changes were not detailed.

In the competitive landscape of AI-powered chat applications, this ban notably contrasts with Microsoft's approach to other applications. For instance, Perplexity, a competitor in the chat app sphere, remains available on the Windows app store, suggesting a selective approach by Microsoft concerning security and competitive strategy. On a related note, apps developed by Microsoft’s major competitor, Google, such as the Chrome browser and Google's chatbot Gemini, were not present on a search of the Windows store.

DeepSeek's presence and policies prompt broader discussions about international data privacy, security, and the influence of geopolitical factors on technology companies' decisions. Microsoft's handling of DeepSeek underscores the complexities tech giants face in balancing innovation, security, and geopolitical risk while maintaining competitive viability in a rapidly evolving tech landscape.

Sources: Commerce Senate, TechCrunch, Microsoft