DeepSeek's newly released AI model, R1-0528, has come under scrutiny for its handling of controversial topics, especially regarding the Chinese government. A developer identifying as 'xlr8harder' noted that the model is less willing to address contentious discussions, particularly those relating to human rights violations in China. For instance, while the model acknowledges the existence of human rights abuses at the Xinjiang internment camps, it avoids direct criticism of the Chinese government's actions. The developer pointed out that this model appears to be the most censored version concerning discussions critical of China, stating that DeepSeek is regressing when it comes to free speech. Although the new version claims improvements in reasoning and inference, it has been criticized for contradictory responses. Observers have highlighted that while the model can cite the camps as human rights violations, it refuses to engage further when provoked for a direct opinion.

Source 🔗