What changed
A study by Vals AI indicates that Chinese AI models do not uniformly align with Chinese government positions on political questions. The research, which tested responses from Chinese, American, and other language models, found that the political leanings of these models vary significantly. The study measured alignment with Chinese government positions on a scale from −1 to +1, where positive values indicate greater agreement with official positions or a pro-government stance, and negative values indicate more critical responses.
What this means in practice
The study found that responses from Chinese models were more pro-government when prompted in Simplified Chinese compared to English, Russian, or Spanish. This effect was more pronounced in Chinese models than in American ones. Additionally, the language used in system prompts influenced responses, with models responding more pro-government when told they were being used internally by a company on mainland China. Some models, like GLM-5.3 Flash, explicitly confirmed documented human rights violations in China when asked, while others, such as Kimi K3, denied the existence of Uyghur concentration camps in Xinjiang, calling them vocational training centers.
Limitations and availability
The study's findings are based on targeted test questions, which may not represent typical user prompts. Refusals to answer were counted as neutral in the scoring, meaning a model that blocks all questions could appear neutral on the scale. Many Chinese models were tested through third-party providers rather than official services, which may apply additional content filters. The study also notes that political tendencies were harder to detect in realistic work tasks than in targeted questions. The results do not necessarily reflect publicly available chatbots, as official offerings may have additional filters.
