From Financial Times: 2024-05-22 05:31:59

Beijing has trained a chatbot on President Xi Jinping’s philosophy to control AI use in China. The new large language model can answer questions, create reports, summarize information and translate Chinese to English. Officials have published books and required studying Xi’s political philosophy, with generative AI providers required to embody socialist values.

Chinese tech giants like Baidu and Alibaba strictly control generative AI content related to Xi or sensitive issues. To help developers comply with speech norms, a database of 100mn entries of “high-quality and trustworthy data” was released for model training. The data draws heavily from government regulations, policy documents, and state media reports.

The Cyber Security Association of China, aligned with CAC, aims to control generative AI output to align with China’s speech norms. Model developers face challenges due to sparse Chinese language data sets, with most groups training on English data as well. Generative AI cannot contain content that subverts state power, says CAC.



Read more at Financial Times: China’s latest answer to OpenAI is ‘Chat Xi PT’