Alibaba to Ban Claude Code in the Workplace Over Alleged Backdoor Risks

Alibaba plans to cut off employee access to Claude Code, the AI coding tool built by Anthropic, in workplace settings starting July 10, according to Reuters, which cited a source familiar with the matter.

The company has reportedly classified the tool as high risk internally over concerns that embedded backdoors could give outsiders access to its systems.

Alibaba did not immediately respond to Reuters’ request for comment, and the move was first reported by Chinese financial outlet Yicai. The ban marks the latest sign of strain in the increasingly adversarial relationship between Alibaba and Anthropic.

A Dispute Rooted in Accusations of Data Theft

The restriction on Claude Code comes months after Anthropic accused Alibaba of carrying out what it called the largest AI distillation attack in history. 

The San Francisco-based AI firm alleged that Alibaba used about 25,000 fraudulent accounts to bombard Claude with millions of requests and use the responses to train its Qwen models, calling the campaign “brazen” and “illicit.” 

Alibaba has since continued expanding its Qwen family, including agentic coding models designed to compete with Claude Code. 

Whether the reported backdoor security concerns reflect a genuine security finding or a response to the earlier dispute remains unclear, as Alibaba has not disclosed the evidence behind the workplace restriction. 

Part of a Wider Pattern of Corporate Restrictions

Alibaba isn’t alone in pulling back from Anthropic’s coding tools. Meta placed strict limits on both Claude Code and OpenAI’s Codex for its engineers just last week, the same month when Anthropic doubled its Claude Code limits.

The social media giant cited worries that AI-generated outputs could be harvested to train competing models. 

Goldman Sachs disabled Claude access for its Hong Kong-based bankers back in April, and JPMorgan Chase followed with a similar restriction in June, both banks pointing to licensing terms and data-security concerns. 

Other Chinese technology firms have also moved away from Anthropic’s tools. ByteDance shut down Claude models on its Singapore-based coding app, Trae, in 2025 after Anthropic tightened its terms of service to cut access from Chinese-owned firms.

While Ant Group had its employees’ access to Claude curtailed after previously offering corporate accounts through a Singapore intranet.

Market Reaction Remains Muted as AI Rivalry Deepens 

As TipRanks notes, Alibaba’s stock has remained under pressure, hovering around $96 following recent reports of insider selling and unrelated legal issues, and the Claude Code restriction has done little to change that trend. 

Beyond the market reaction, the dispute reflects a broader escalation in U.S. and Chinese AI competition. 

Allegations of model distillation and intellectual property misuse have already heightened scrutiny of Chinese AI firms, while enterprise security concerns are increasingly shaping how companies evaluate and restrict rival AI tools. 

As AI competition intensifies, trust and security are becoming as important as model performance.

Source: Alibaba to ban employees from using Anthropic’s coding tool, source says