Microsoft (MS) has announced that it has completely blocked the internal use of DeepSeek, a Chinese artificial intelligence (AI).
According to foreign media outlets including TechCrunch on the 9th, Brad Smith, MS vice chairman and president, noted during a U.S. Senate hearing on the 8th that "employees are not allowed to use the DeepSeek app." DeepSeek is a generative AI chatbot available on desktop and mobile platforms.
MS has also excluded the DeepSeek app from its company app store, citing that DeepSeek stores data in China and filters sensitive topics according to Chinese government censorship criteria.
According to DeepSeek's privacy policy, user data is stored on servers in China, which are subject to the National Intelligence Law of China and mandate cooperation with government intelligence agencies. Additionally, DeepSeek provides limited responses to topics that are sensitive to the Chinese government.
This is the first time MS has publicly expressed concerns about DeepSeek. Some interpret this as an intention by MS to restrict DeepSeek, which competes with its own AI service, "Copilot." However, MS allows some other competing apps, and generative AI chatbots like Perplexity can be searched and installed from the MS Windows app store.
While MS has prohibited the use of the DeepSeek app, it has integrated DeepSeek's open-source model "R1" into its cloud platform Azure. The model was described as having undergone "strict red team evaluations and safety checks."
Due to its open-source nature, anyone can download the DeepSeek model and run it on their own servers, which reduces security risks since data is not sent to China. However, risks inherent to the model itself, such as propaganda dissemination or code security issues, still exist.
Vice Chairman Brad Smith stated, "We were able to enter the internal model of DeepSeek and eliminate harmful side effects," adding that, "It is difficult to disclose more specific measures."