Alibaba Group has barred employees from using Anthropic's AI coding tool "Claude Code" in the workplace, according to multiple news outlets citing sources. The ban took effect on July 10, 2026 for office environments, with instructions to remove the related software from work computers.
July 3, 2026 · Alibaba Group
Alibaba Bans Claude Code as "High-Risk Software"
After security researchers found a hidden feature that fingerprinted Chinese users, Alibaba classified Anthropic's coding tool as high-risk — ordering staff onto its in-house platform Qoder from July 10, 2026.
147
domains tied to Chinese firms the tool matched against
25,000
fake accounts Anthropic alleged were used to distill its models
28M
interactions generated over six weeks, per the letter to senators
The tracking feature — its lifespan in the tool
v2.1.91
Apr 2, 2026
→
active
fingerprinting
→
v2.1.197
removed
How the feature reportedly worked
Inspect environment
Time zone & proxy settings read locally
→
Match domains
Cross-referenced vs 147 China-linked domains
→
Insert markers
Hidden Unicode markers added to prompts sent upstream
Developers: "spyware / backdoor"
Agentic tools access local files, terminals, projects
Environment inspection seen as a privacy & security risk
Fear of broader erosion of global trust
Anthropic's position
An experiment begun March 2026 to curb account abuse
Aimed at unauthorized resellers & model distillation
Says stronger mitigations added; feature was to be removed
The takeaway: the episode turns "supply-chain trust" — trust in how an AI coding tool actually behaves — into a first-order concern, with growing calls for OS-level auditing and transparency in adoption decisions.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…