DeepSeek banned by Microsoft — what's wrong with the chatbot
Microsoft employees are not allowed to install and use the DeepSeek app on their work devices. It was announced during the proceedings in the United States Senate by Brad Smith, Vice Chairman and President of the corporation. The ban, according to him, is related to the risk of data storage in China and the possible influence of "Chinese propaganda" on chatbot responses.
It was reported by TechCrunch.
Why does Microsoft consider DeepSeek to be dangerous?
Smith clarified that DeepSeek is not available on the Microsoft Store either — the corporation deliberately does not offer it for download. While several organisations and even governments have already imposed restrictions on DeepSeek, this is the first time Microsoft has publicly confirmed such a move.
DeepSeek's Privacy Policy states that user data is stored on Chinese servers and is subject to local laws that oblige companies to cooperate with intelligence. In addition, the service actively censors topics that are sensitive to the Chinese authorities.
Despite the harsh assessment, Microsoft still offered the DeepSeek R1 model on its Azure cloud platform immediately after it went viral. The company explains that open source allows customers to deploy the model locally without transferring data to China, although this does not exclude other risks, from disinformation to dangerous software code.
During the proceedings, Smith noted that Microsoft engineers had "changed" the model's internal mechanisms to eliminate "harmful side effects", but the company did not disclose the details of these changes. Earlier in the blog post about the Azure launch, it was reported that DeepSeek had undergone "thorough testing and red teaming" before being released in the cloud.
Despite the competition with its own Copilot, Microsoft does not block all third-party chatbots: for example, Perplexity is available in the Windows Store. At the same time, the store does not offer Google applications, including the Chrome browser or the Gemini chatbot.
As a reminder, Microsoft has significantly expanded the functionality of Copilot Studio by adding the tool called Computer Use. It allows corporate AI agents to act in the computer interface almost like human employees: press buttons, select menu items, and enter text in the required fields.
We also wrote that the company actively integrates generative technologies into the software development process. According to CEO Satya Nadella, AI is currently responsible for creating 20% to 30% of the code in Microsoft's internal repositories.
Read Novyny.LIVE!