Alibaba restricts Claude Code access citing security risks

The restriction arrives as major financial institutions simultaneously retreat from Anthropic’s tools. Last month, JPMorgan and Goldman Sachs curbed employee access to Claude models in Hong Kong, citing inconsistencies between the platform’s licensing terms and regional usage policies. While Anthropic maintains that its models were never officially supported in that jurisdiction, these incidents highlight a broader trend of enterprise caution regarding the integration of generative AI into sensitive workflows.

This development follows a period of regulatory friction for Anthropic. On July 1, the company reinstated its Claude Fable 5 and Mythos 5 models after a brief suspension mandated by U.S. export controls. To appease government concerns over cybersecurity, Anthropic implemented new classifiers intended to detect and block potential jailbreak attempts. Despite these updates, Alibaba’s move suggests that internal security teams remain wary of the underlying architecture of these AI systems, prioritizing proprietary data protection over the efficiency gains offered by external coding assistants.