Alibaba has directed its employees to stop using Anthropic’s Claude Code after the AI tool began blocking connections from China. This escalation follows allegations that Alibaba ran large-scale data extraction campaigns, using around 25,000 fake accounts to train its own models on Claude’s outputs. Anthropic claims this constitutes unauthorized use of proprietary training data, raising concerns about intellectual property and ethical boundaries in AI development. The dispute highlights growing tensions between global tech leaders over access to AI tools and the integrity of model training processes.
The conflict underscores broader challenges in the international AI landscape, where companies face difficult decisions about collaboration versus protection of their technologies. As nations seek to strengthen domestic AI capabilities, restrictions on foreign tools are becoming more common. This situation reflects a larger shift toward national control over AI infrastructure and data, potentially reshaping how global firms develop and deploy artificial intelligence systems in the future.