>>13655835
As part of its broader information strategy, China has been actively developing and deploying artificial intelligence technologies, notably the DeepSeek model. Its free availability and high performance have contributed to its rapid adoption across the post-Soviet states and globally, including in the United States. Given that DeepSeek was developed by a Chinese company, there are legitimate concerns that its architecture and training algorithms may be designed to subtly promote Chinese state interests to users worldwide.
For non-political, everyday applications, DeepSeek likely poses minimal risk. However, when engaged on politically sensitive topics, the model has been shown to exhibit bias, frequently reflecting the official positions of the Chinese government. Research has documented DeepSeek's tendency to suppress information regarding government accountability and civic mobilization, and to outright refuse responses on topics such as democracy and human rights. Furthermore, studies have identified that DeepSeek contains security vulnerabilities that can be exploited through prompt engineering, effectively bypassing its built-in safety filters and potentially allowing users to extract information promoting fascist or anti-China ideologies