Alibaba Bans Claude Code Over Backdoor Fears
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Alibaba banned internal use of Anthropic's Claude Code and ordered employees to remove all Claude models from work computers, citing alleged backdoor risks (The Information, Reuters)
- The ban, set to take effect July 10, directs employees to Alibaba's own AI coding platform Qoder as a replacement (per X posts from Poe Zhao and Andrew Curran)
- Anthropic is separately moving to close workarounds that let Chinese companies including Ant Group access its models via cloud providers and overseas subsidiaries (FT)
- Coverage converges on a security framing: Reuters cites "backdoor risks," SCMP names "spyware concerns," The Next Web reports "tracking Chinese users with hidden code," and Cyber Security News points to "embedded backdoor risks"
- Kevin Xu wrote on X that a broader watchlist of potential next targets is "LONG," naming Baidu, Kuaishou, ByteDance, DeepSeek, and Moonshot alongside Alibaba
- The Decoder frames the broader pattern as "bans on both sides of the Pacific," pointing to reciprocal restriction between Anthropic and Chinese firms rather than a one-sided crackdown
Why it matters: The incident formalizes a two-way AI firewall between Anthropic and Chinese tech firms. With Alibaba—one of China's largest tech employers—banning Claude internally while Anthropic simultaneously severs overseas-subsidiary workarounds, a de facto split hardens between Chinese and Western AI development stacks; domestic alternatives including Qoder, DeepSeek, and Moonshot are positioned to absorb the displaced usage. The move also implicitly confirms Alibaba had been tolerating unauthorized Claude use, since Anthropic does not officially serve Chinese users.
