China has issued a security warning on Anthropic\'s AI coding tool Claude Code, raising the likelihood that Chinese developers will move faster to domestic alternatives. The National Vulnerability Database said some versions contain a security \"backdoor\" and could transmit location and identity data without user consent. Anthropic said access from China had long been prohibited under its policy. Local observers expect demand to rise for tools from ByteDance, Alibaba, Tencent Cloud and ZhipuAI.

Concern is growing within China’s AI industry about its competitiveness in large language models, with some firms seeking a breakthrough in video-generation AI. Liu Wei, who led Tencent’s Hunyuan generative AI model team, cited a lack of paradigm innovation and said Chinese companies are copying U.S. approaches in core technologies. He estimated the U.S.-China gap at at least 3 months and warned it could widen to 6 months this year.