Alibaba is reportedly telling employees to stop using Anthropic’s AI coding assistant, Claude Code, starting July 10, following an internal review that classified the software as high risk. The decision comes after recent reports that the tool included experimental code designed to help identify certain users in China, adding another layer to the ongoing tensions between U.S. and Chinese AI companies.
Anthropic has said the code was part of an experiment aimed at preventing unauthorized resellers and protecting its models from distillation, and that it has already been removed.
Even so, Alibaba is reportedly directing employees to switch to its own AI coding platform, Qoder.
Why It Matters: This shows how AI companies are trying to protect their technology while navigating different rules, security concerns, and growing competition around the world. As U.S. and Chinese companies continue to build more of their own AI ecosystems, decisions like this are becoming more common.
- Alibaba is moving employees to its own AI coding platform. According to multiple reports, the company has classified Claude Code as high-risk software and instructed employees to stop using it beginning July 10. Instead, staff are being directed to use Qoder, Alibaba’s in-house coding assistant. Some reports also suggest the guidance extends to other Anthropic products used in the workplace.
- The decision follows claims about hidden detection code. A developer who reverse-engineered Claude Code reported finding code that could identify signs a user was located in China or connected to certain Chinese AI organizations. The report also claimed that this information was communicated back to Anthropic in subtle ways that weren’t visible to users, raising questions about transparency and how the software handled account verification.
- Anthropic says the feature was an experiment that has already been removed. Claude Code engineer Thariq Shihipar explained that the code was introduced as part of an effort to reduce abuse by unauthorized resellers and help protect against model distillation, where one AI system is trained using another model’s outputs. He said the team had already planned to remove the experiment and merged changes to do so shortly after it became public.
- The disagreement comes after weeks of rising tensions between the two companies. In June, Anthropic accused operators linked to Alibaba’s Qwen AI lab of using thousands of accounts to generate millions of Claude interactions in what it described as a large-scale model distillation effort. Alibaba has denied wrongdoing, but the dispute has added to an already complicated relationship between the companies.
- The bigger picture goes beyond Alibaba and Anthropic. AI companies are increasingly balancing security, intellectual property protection, and international regulations while expanding their products around the world. As access restrictions and regional policies continue to evolve, companies on both sides are placing greater emphasis on building and relying on their own AI ecosystems.
Go Deeper -> Alibaba reportedly bans employees from using Claude Code – TechCrunch
Trusted insights for technology leaders
Our readers are CIOs, CTOs, and senior IT executives who rely on The National CIO Review for smart, curated takes on the trends shaping the enterprise, from GenAI to cybersecurity and beyond.
Subscribe to our 4x a week newsletter to keep up with the insights that matter.
