Deepseek.ai servers

Do the Deepseek.ai servers source from China under government policies?

DeepSeek, a Chinese AI company, has released AI models that appear to be influenced by Chinese government policies and regulations. The company's DeepSeek R1 model, launched on January 20, 2025, has been observed to exhibit censorship tendencies, particularly when addressing topics sensitive to the Chinese government1.

Users have reported that DeepSeek R1 avoids answering questions about controversial subjects, such as changes in Hong Kong textbooks or large public gatherings. The model also self-censors when discussing topics that might contradict Chinese government narratives1.

This behavior is likely a result of China's AI regulations, implemented in 2023. These regulations require AI providers to adhere to "core values of socialism" and prohibit content that could be seen as subverting state power or undermining national unity1.

While DeepSeek has released open-source models under an MIT license, allowing free access with minimal restrictions, the underlying influence of Chinese government policies on the AI's responses is evident1. This suggests that DeepSeek's servers and AI development are indeed subject to Chinese government oversight and policies.