Alibaba bans employees from using Claude Code due to high risk

What happened?

According to an exclusive TechCrunch report from March 10, 2025, Alibaba has internally classified Claude Code, the AI-assisted coding tool developed by Anthropic, as high-risk software, prohibiting its use by all employees. The decision, not yet officially confirmed by the company, adds to previous restrictions on other AI tools such as ChatGPT (blocked in 2023) and GitHub Copilot (restricted in 2024). Claude Code, launched in January 2025, allows developers to generate, debug, and refactor code through natural language commands, with advanced contextual understanding capabilities that surpass competitors like GitHub Copilot or Amazon CodeWhisperer.

Why is it important?

This measure is not an isolated case. China has intensified its control over foreign AI technologies, especially those developed by US companies. Claude Code, running locally in employees' development environments, could expose proprietary code to Anthropic's external servers, posing security and data leakage risks. Alibaba, with interests in cloud computing (Alibaba Cloud, holding a 40% share of the Chinese cloud market), e-commerce (Taobao, Tmall), and financial services (Ant Group), has much to lose if its intellectual property is exposed. Additionally, the Chinese regulatory context is increasingly strict: in 2024, the Cyberspace Administration of China (CAC) issued guidelines requiring companies to assess the risks of foreign AI tools, and Alibaba may be anticipating potential sanctions.

Historical context

In 2023, Alibaba blocked access to ChatGPT on its internal networks, citing security concerns. In 2024, it restricted GitHub Copilot following Microsoft's acquisition of GitHub. The ban on Claude Code follows the same line, but with an additional nuance: Anthropic's tool is perceived as more powerful, with reasoning and code generation capabilities that could be considered "strategic." Tensions between the US and China over technological supremacy have led both nations to restrict access to certain technologies. For example, in October 2024, the US tightened export restrictions on AI chips to China, affecting companies like Huawei and Alibaba. In response, China has fostered the development of local alternatives, such as Alibaba's Qwen model, which directly competes with Claude and ChatGPT.

Consequences for the ecosystem

• For Alibaba: The company will need to seek local alternatives, such as CodeGeeX from Zhipu AI, or its own code assistant based on Qwen. This could slow down its software development productivity, as local tools have yet to match the maturity of Claude Code. However, Alibaba has an incentive to accelerate its own models and reduce foreign dependence.
• For Anthropic: Losing a potential client as large as Alibaba (with over 200,000 employees) is a setback, but the US startup will likely focus on more open markets like the US, Europe, and Japan. Anthropic has raised over $7 billion to date, with a valuation exceeding $60 billion, giving it room to absorb this impact.
• For the global market: The fragmentation of the AI ecosystem is accelerating. Multinational companies like Apple, which operates in China, may be forced to maintain two sets of tools: one for China (with local models) and another for the rest of the world. This increases development costs and operational complexity. According to a February 2025 Gartner report, 30% of large global companies have already adopted dual AI strategies to comply with local regulations.

What should readers know?

Developers working for companies with a presence in China should prepare for similar restrictions. Open-source AI tools, such as Meta's Code Llama or Hugging Face's StarCoder, could gain traction in the Asian country, as they allow full control over data. Moreover, Alibaba's decision sends a signal to other major Chinese corporations, such as Tencent, Baidu, and ByteDance, which could follow suit. In fact, internal sources at Tencent cited by Reuters suggest the company is evaluating similar restrictions for Claude Code. On the other hand, the "high-risk" classification could inspire regulators in other countries, such as the EU, to consider risk labels for AI tools in the workplace.

Alibaba's classification of Claude Code as "high risk" is a reminder that AI regulation comes not only from governments but also from companies themselves, acting as guardians of their intellectual property.

Expert analysis

According to industry analysts, this measure could have a domino effect. If other Chinese companies imitate Alibaba, Anthropic could lose a significant portion of the Asian market, which represents approximately 25% of its potential revenue according to PitchBook estimates. However, some see an opportunity for Chinese AI startups, such as Zhipu AI or Baidu (with its Ernie model), to develop safer code assistants tailored to local regulations. Dr. Li Zhang, a computer science professor at Peking University, commented in a South China Morning Post article: "Alibaba's decision is strategic: it protects its competitive advantage and fosters the local AI ecosystem. In the long run, this could strengthen China's technological sovereignty." On the other hand, cybersecurity experts like Kaspersky point out that the risk of data leakage is real: in 2024, a similar incident with a closed-source AI tool exposed credentials of a Fortune 500 company.

Conclusion

Alibaba's ban on Claude Code is a symptom of the growing technological distrust between China and the US. For industry professionals, the lesson is clear: geopolitics is reshaping the generative AI landscape, and adaptability will be key to survival. Companies must evaluate not only the functionality of AI tools but also their origin, data storage, and regulatory compliance. In an increasingly fragmented world, the ability to quickly switch between AI providers could become a decisive competitive advantage.