Microsoft Bans Employees from Using DeepSeek App Over Data Security and Propaganda Concerns

Microsoft has forbidden its employees from using the DeepSeek app due to data security and propaganda concerns, according to Brad Smith, the company's vice chairman and president, during a Senate hearing today. "At Microsoft, we do not allow our employees to use the DeepSeek app," Smith stated, emphasizing that the ban applies to both desktop and mobile versions of the application. Additionally, Microsoft has refrained from adding DeepSeek to its app store, citing the same concerns. While various organizations and even countries have implemented restrictions on DeepSeek, this marks the first time Microsoft has publicly disclosed such a ban. Smith highlighted two primary reasons for the policy: the risk of user data being stored in China and the potential influence of Chinese propaganda on DeepSeek's responses. According to DeepSeek’s privacy policy, user data is stored on Chinese servers, which are subject to Chinese law. This legal framework requires companies to cooperate with the Chinese intelligence community, raising significant data security issues. Furthermore, DeepSeek enforces strict censorship on topics deemed sensitive by the Chinese government. Despite these criticisms, Microsoft made the R1 version of DeepSeek's AI model available on its Azure cloud service shortly after the app went viral earlier this year. However, this is distinct from offering the chatbot app directly. Since DeepSeek is open-source, users can download the model, host it on their servers, and serve their clients without transmitting data back to China. Nonetheless, this approach does not eliminate other risks, such as the model's potential to spread propaganda or generate insecure code. During the hearing, Smith revealed that Microsoft had conducted internal modifications to the DeepSeek AI model to mitigate "harmful side effects." The company, however, declined to provide detailed explanations, referring TechCrunch to Smith’s testimony. In the initial launch of DeepSeek on Azure, Microsoft noted that the AI model underwent "rigorous red teaming and safety evaluations" before it was made available. This process aimed to identify and address potential security and ethical issues. It's worth noting that DeepSeek’s app competes directly with Microsoft’s Copilot, an internet search chat app. Despite this competition, Microsoft has not implemented a blanket ban on all chat-based competitors in its Windows app store. For example, Perplexity, another chat app, remains available. On the other hand, apps from Microsoft’s rival, Google, including the Chrome browser and Google’s chatbot Gemini, were notably absent from the Windows webstore search results. By taking these steps, Microsoft aims to balance the benefits of innovative AI tools with the necessity of protecting user data and maintaining ethical standards. The company’s decision underscores the ongoing concerns about data privacy and security in the rapidly evolving landscape of artificial intelligence applications.