Alibaba Claude Code Ban: What the Internal Notice Says
Alibaba has told its staff to stop using Anthropic's Claude Code at work, SCMP reported on July 3, 2026. Claude Code is Anthropic's AI coding agent for developers.
For the model that powers it, see What Is Claude AI? Free Plan, Models, vs ChatGPT. It covers the plans and models in one place.
Timeline of the Claude Code ban, as reported (2026)
- MarchAnthropic launches the experiment that Shihipar later describes (his account)
- Jul 2 (Thu)Alibaba's internal notice, which SCMP says it saw
- Jul 3SCMP publishes its report of the ban
- Jul 4TechCrunch reports the ban, citing multiple reports
- Jul 10Start date the notice sets for barring staff from using Claude Code in the office
The notice and the July 10 start date
From July 10, no Alibaba employee may use Claude Code in the office, according to the notice. SCMP says it saw the notice, which went out on Thursday, July 2. When TechCrunch covered it on July 4, the ban had not yet taken effect, and its report uses the future tense.
"The Chinese technology giant said all staff members would be prohibited from using Claude Code in the office starting from July 10." — SCMP's report on the noticeView official source →
Why Alibaba calls Claude Code high-risk software
After what it calls a comprehensive evaluation, Alibaba added Claude Code to its list of high-risk software with security vulnerabilities. The reason given is a back-door risk. That is Alibaba's judgment, relayed by SCMP and not confirmed by anyone else.
"As Claude Code was recently discovered to carry back-door risks, after comprehensive evaluation, Claude Code has now been added to a list of high-risk software with security vulnerabilities," — wording reported as Alibaba's internal noticeView official source →
Qoder as the replacement tool
Staff are reportedly being moved to Qoder, Alibaba's own coding tool. TechCrunch reports the move together with the high-risk label Alibaba gave Claude Code.
"Nonetheless, Alibaba has reportedly classified Claude Code as high-risk software and is instructing employees to use the company’s Qoder tool instead." — TechCrunchView official source →
Claude Code Backdoor Risk: The Hidden Code Behind the Ban
Reports link the ban to code hidden inside Claude Code. SCMP cites people familiar with the matter, and TechCrunch points to a recent Reddit post. Anthropic's own account calls that code an experiment, not a back door.
What the code could reportedly detect
According to SCMP, Anthropic had added code to Claude Code that could quietly check whether a user was in China. SCMP also says, citing "people familiar with the matter," that the code could also tell whether a user was affiliated with a Chinese AI lab.
In SCMP's account, this is the code Alibaba's notice refers to. TechCrunch adds that a recent Reddit post flagged a version of Claude Code able to identify Chinese users without telling them.
"Anthropic’s earlier move to embed code in its Claude Code coding agent platform that could secretly track whether a user was based in China or affiliated with a Chinese AI lab, according to people familiar with the matter." — SCMPView official source →
Anthropic's explanation: an experiment from March
Anthropic's Thariq Shihipar wrote on X that the code was an experiment started in March. In the post TechCrunch quotes, he says its goal was to stop unauthorized resellers from abusing accounts and to guard against distillation. Stronger protections have shipped since then, he added, and the team had meant to remove the experiment.
"an experiment we launched in March that was meant to prevent account abuse from unauthorized resellers and protect against distillation." ／ "The team has landed stronger mitigations since then and we’ve actually been meaning to take this down for a while," — TechCrunch quoting ShihiparView official source →
Anthropic China Access Limits and Distillation Concerns
Chinese companies were already shut out of Anthropic's models, and TechCrunch, citing a Reddit post, treats the hidden code as part of that wider effort. All of this background comes from TechCrunch's report.
What Anthropic already prohibits
Neither Chinese companies nor the foreign entities they own may use Anthropic's models. TechCrunch adds that Anthropic has reportedly been closing gaps that still let Chinese users get to Claude. A Reddit post it cites says part of that work involved a Claude Code version that could identify those users without their knowledge.
"Anthropic already prohibits Chinese companies, as well as foreign entities owned by those companies, from using its models." — TechCrunchView official source →
What distillation means here
Distillation, as TechCrunch defines it, is training an AI model on what another model outputs. Shihipar listed it as one of the two things the experiment was supposed to stop. AI Distillation Attacks in PLA-Linked Research covers the technique in more detail through a separate case.
Alibaba's stated reason is the hidden code
SCMP puts the hidden code, not distillation, behind Alibaba's notice. The words quoted from the notice name back-door risks only and say nothing about distillation or a dispute with Anthropic; the link to the hidden code is SCMP's own summary. Distillation comes into the story only through Shihipar's reason for the code.
"citing security risks related to the US artificial intelligence firm’s previous use of hidden code to track Chinese users" — SCMP's own summary, not the noticeView official source →
Is Claude Code Safe? What the Ban Means Beyond Alibaba
None of the reports include an independent test of Claude Code's safety; the ban rests on Alibaba's own assessment and applies only to its staff at work. Anthropic, for its part, calls the code behind that assessment a March experiment it had planned to remove.
What the notice covers: Alibaba staff at work
What is reported is a rule for Alibaba employees using Claude Code on the job, not a warning for other users. SCMP's wording limits it to use for work and in the office. Anthropic's ban on Chinese companies is a separate policy that TechCrunch says came earlier.
"Alibaba Group Holding has banned its employees from using Anthropic’s Claude Code for work" — SCMPView official source →
Two claims: back-door risk and an experiment
The reports carry two claims side by side: Alibaba's back-door risk and Anthropic's March experiment. Alibaba's is in a notice SCMP says it saw, and Anthropic's is in an X post TechCrunch quotes. One is made by the company issuing the ban and the other by the company whose code set it off.
Who made each claim, and who reported it
Every claim has its own source and reached the public through its own outlet, so they do not carry equal weight. The table lists who said each one and who reported it.
| Claim | Said by | Reported by |
|---|---|---|
| Claude Code is on a high-risk list; staff to be barred from July 10 | Alibaba's internal notice | SCMP (saw the notice); TechCrunch (multiple reports) |
| Hidden code could tell if a user was in China | People familiar with the matter; a Reddit post | SCMP; TechCrunch |
| The code was a March experiment against abuse and distillation | Thariq Shihipar on X | TechCrunch |
| Staff are told to use Qoder | Alibaba, reportedly | TechCrunch |
Going by the July 3 and 4 reports, Alibaba will keep its own staff off Claude Code from July 10 because of back-door risks. Anthropic says the code behind that worry was a March experiment aimed at account abuse and distillation.
FAQ
- Why did Alibaba ban Claude Code?
- According to an internal notice that SCMP says it saw, Alibaba added Claude Code to a list of high-risk software with security vulnerabilities after a comprehensive evaluation, citing back-door risks. SCMP ties this to Anthropic's earlier hidden code that could track whether users were in China. Anthropic's Thariq Shihipar has described that code as a March experiment.
- What was the code that could identify China-based users?
- SCMP, citing people familiar with the matter, says Anthropic embedded code in Claude Code that could secretly track whether a user was in China or tied to a Chinese AI lab. Anthropic's Thariq Shihipar said on X, as TechCrunch quotes him, that it was an experiment launched in March against account abuse and distillation. He added that the team had been meaning to take it down for a while.
- Does the Alibaba ban affect other Claude Code users?
- The reported ban is a rule for Alibaba's own staff. SCMP says all of them would be barred from using Claude Code in the office from July 10. Anthropic's restriction on Chinese companies and the foreign entities they own is a separate, earlier policy, according to TechCrunch.
Related articles
Claude Code rule blocks recreating existing gamesA server-side rule stops Claude Code on Opus 5.5 from recreating existing games. Anthropic confirmed it on X, and a wargame studio says paid work stopped.
Claude Code hooks: 5 types and the node spawn costClaude Code hooks come in five types, from command to agent. As command hooks, one Edit launched 17 node processes, and hooks drove 47–87% of node launches.
Claude Code Auto Mode Is Now the Default SettingClaude Code auto mode is now the default on Pro, Max and Team. Learn what the classifier blocks, how to switch modes, and how to disable it.
Claude Code Token Usage: 33K vs OpenCode's 7KClaude Code token usage starts near 33,000 before your prompt, 4.7x OpenCode. CLAUDE.md and MCP servers push it to 75,000, but requests set the total.