Alibaba bans internal use of Anthropic's Claude

Alibaba has ordered employees to stop using Anthropic's Claude family of AI models and related coding tools, marking a sharp reversal from its previous policy of encouraging staff members to use leading overseas AI products, sources confirmed to China Daily.

According to two sources with direct knowledge of the decision, the internal directive, which takes effect on July 10, requires employees to uninstall Anthropic products, including Claude Sonnet, Claude Opus, Claude Fable and the company's AI coding agent Claude Code.

Since early this year, the company had actively encouraged employees to adopt AI tools by providing free quotas for its own models while reimbursing staff members for the use of third-party services alongside offering free quotas for its own models, the sources said.

Employees were allowed to use Anthropic's Claude, OpenAI's GPT models and Google's Gemini, with some teams spending hundreds of dollars a week on external AI services, they added.

The dispute has also been fueled by allegations circulating among Chinese developers this week that Claude Code contained hidden mechanisms for identifying users linked to China.

A Reddit user who said they had reverse-engineered the software alleged it used steganographic techniques — such as subtly changing date formats and substituting visually indistinguishable punctuation characters in system prompts — to secretly mark Chinese users and transmit those identifiers back to Anthropic.