Alibaba Bars Use of Anthropic's Claude Code Amid Data Security Concerns

Alibaba is reportedly putting the brakes on its employees' use of Anthropic's coding tool, Claude Code, following scrutiny over the software's potential to track users connected to China. An insider familiar with the matter revealed that this move is part of a growing rift between the two tech giants.

Anthropic, a U.S.-based AI firm, has accused Alibaba of attempting to replicate its AI models through a tactic known as "distillation," which involves training a weaker model on the output of a stronger one. This technique could catapult China's AI capabilities considerably closer to Anthropic's advanced offerings, the company has warned in confidential communications.

Claude Code functions as an AI assistant tailored for software developers and has gained traction among Chinese programmers despite Anthropic's efforts to restrict access within China. However, Alibaba employees are now being directed to shift towards the firm's internal coding platform, Qoder.

Technical revelations surfaced recently showing that Claude Code incorporates features to collect environment data such as time zones and proxy information, as well as embedding subtle markers in server requests. This tracking mechanism, reportedly initiated in March, aims to prevent unauthorized resale of the software and guard against model distillation.

Enforcing regional access restrictions has proven challenging because users inside China can route data through servers abroad, particularly in the U.S., making detection tricky. While individual users might navigate around these blocks, companies like Alibaba face stricter legal and compliance considerations, prompting more cautious policies.

This episode underscores a broader picture: American AI developers are beefing up defenses against unapproved use, while Chinese cloud and AI providers increasingly lean on homegrown and open-source AI models like Alibaba's own Qwen, as well as DeepSeek and Moonshot.

Interestingly, some Chinese AI systems are quietly establishing footholds in the U.S. market, a development that has sparked unease within U.S. industry circles about the implications for technological leadership and security.

While Alibaba has not commented publicly on Anthropic's allegations, this ban on Claude Code echoes deepening competition and mistrust amid the world's biggest players vying for dominance in artificial intelligence.

Looking ahead, the question remains whether these cross-border tech tensions will lead to more fragmentation in AI development or push firms towards new forms of collaboration - a dynamic that will surely unfold over the coming months.