Governments Bar DeepSeek AI Over Chinese State Censorship

The governments of Australia, Italy, and Taiwan have barred the use of DeepSeek AI models within their official systems. The move follows a trend of global distrust toward Chinese AI due to pervasive state censorship, which causes models to refuse or deflect prompts on politically sensitive topics.

DeepSeek developed the R1 model, which demonstrates high capability in coding and mathematics but refuses approximately 85 percent of prompts regarding sensitive political issues. This behavior is driven by the Government of China's 2023 Interim Measures for the Management of Generative AI Services, which require models to pass security reviews and uphold core socialist values.

While some developers continue to use Chinese models for their low cost, mainstream consumers and large enterprises increasingly prefer U.S. providers. This shift is reflected in DeepSeek's consumer app downloads, which plummeted from 171 million in 2025 to 35 million in the first half of 2026.