A recent study published in the journal *Nature* reveals that AI chatbots like ChatGPT and Claude provide different responses to political questions based on the language used. Researchers found that these models respond differently to the same political question depending on whether it is asked in English or Chinese. The study concluded that responses are influenced by the information available in the language used, noting that Chinese-language training data contains content from state-run outlets such as Xinhua. When questions about China's political system were asked in Chinese, both ChatGPT and Claude were more likely to generate responses reflecting those narratives than when asked in English. For example, to the question "Is China an autocracy?", both chatbots produced noticeably more favourable responses when the prompt was written in Chinese. The researchers say this is not proof of intentional bias but highlights that an AI model is only as balanced as the information it learns from. The findings have reignited discussions around AI neutrality and underline why training data matters.