Alibaba bans employees from using Anthropic’s Claude Code after alleged “distillation attack” and security concerns

Alibaba will ban employees from using Anthropic’s Claude Code and other Claude-related products for work purposes starting July 10, according to CNBC. The company also told employees to remove Anthropic models and agent products from internal environments, and to switch to Alibaba’s own AI assistant, Qoder, the report said.

In the internal policy change, Alibaba categorized Anthropic tools as high-risk software, citing concerns that the U.S. company’s systems present “back-door security risks,” CNBC reported. The company’s action means Anthropic’s coding and assistant tools will no longer be available to Alibaba staff for job-related use once the effective date arrives.

CNBC said the ban follows a dispute between the companies in which Anthropic accused Alibaba of carrying out what it called the largest known “distillation attack” on Anthropic to date. Anthropic’s position, as described by CNBC, is that the alleged activity was intended to extract its AI capabilities. The report also said Anthropic had communicated the dispute to U.S. officials, sending a letter in June to the U.S. Senate Committee on Banking, Housing, and Urban Affairs.

Alibaba’s decision also appears consistent with Anthropic’s stated restrictions on usage. CNBC reported that Anthropic’s terms of service bar Chinese companies and other “adversarial nations” from using its models, and that Alibaba’s internal policy references that compliance issue while adding its own security rationale.

The widening rift has also drawn online controversy in China, according to CNBC, which cited posts on platforms including Reddit and GitHub that discussed the use of hidden code intended to detect whether users are located in the country. The report described the dispute as part of a broader wave of scrutiny aimed at how AI tools identify or restrict users, and how third parties might circumvent access controls.

CNBC further reported that the ban comes amid efforts by Anthropic to close workarounds that it said allowed Chinese companies to reach Claude through third countries. That point was attributed to reporting by the Financial Times, which said Anthropic was moving to address loopholes.

For Alibaba employees, the practical change is immediate in internal compliance terms: CNBC said the company required employees to uninstall Anthropic tools and agent products and then use Qoder instead. For Anthropic, the change affects one of the largest potential enterprise workforces in China that had been able to access Claude-related products, while raising the stakes for both companies as they continue to dispute whether security issues were introduced intentionally or through operational attempts to adapt models.

The episode highlights how corporate AI governance is increasingly tied to cybersecurity risk management and international technology restrictions, not just model performance. With Alibaba’s ban set to take effect on July 10, the next measurable step will be whether employees can continue to use approved alternatives internally and whether either company pursues additional formal steps with regulators or lawmakers.