DeepSeek, a Chinese startup, recently gained attention for its open-source AI model, which excels in math and reasoning. However, the model imposes stringent censorship regarding politically sensitive topics like Taiwan and Tiananmen Square. WIRED's testing revealed that users can bypass some censorship on third-party platforms, raising concerns about how such model manipulations could affect the model’s global competitiveness. Should modifications to tackle censorship prove complicated, it may hinder DeepSeek's appeal in the international AI market, especially against less-restrictive US alternatives.
DeepSeek, a Chinese startup, showcases an open-source AI model with strong reasoning capabilities but strict censorship on sensitive topics like Taiwan and Tiananmen.
Censorship in DeepSeek's model can be circumvented on third-party platforms, raising questions about the impact on the competitiveness of Chinese AI globally.
While initial perceptions of censorship issues appear manageable, underlying biases in AI data training may complicate modifications for users seeking unrestricted access.
Response restrictions based on government regulations may lessen the appeal of DeepSeek compared to its US counterparts in a highly competitive market.
Collection
[
|
...
]