Alibaba has reportedly decided to block employees from using Anthropic’s Claude Code in its workplace environments, citing alleged security concerns related to potential embedded backdoors.
The restriction, which is expected to take effect on July 10, marks another setback for Anthropic as large enterprises continue to scrutinize the security, licensing, and regulatory compliance of generative AI tools.
While Alibaba has not publicly confirmed the decision, the reported move highlights the growing focus on AI security within major technology companies.
Alibaba reportedly restricts Claude Code
According to a source familiar with the matter, Alibaba will prohibit employees from using Claude Code across its internal work environments beginning July 10.
The reported decision was based on concerns over alleged security risks involving embedded backdoors within the AI coding assistant.
However, no technical details have been released regarding the alleged vulnerabilities, and Alibaba has yet to issue an official statement explaining the scope of the restriction or providing evidence to support the claims.
As of now, Anthropic has also not publicly responded to the reported ban.
Another enterprise setback for Anthropic
The reported Alibaba restriction follows several recent challenges for Anthropic’s enterprise adoption.
In June, the Financial Times reported that JPMorgan removed Claude models from its list of approved large language models for employees in Hong Kong.
The decision was reportedly linked to Anthropic’s licensing terms, which restricted where the company’s AI models could legally be used.
Around the same time, Goldman Sachs introduced similar limitations after determining that Anthropic’s terms of service excluded usage across Greater China, including Hong Kong.
Anthropic later clarified that Claude had never been officially supported in Hong Kong, while JPMorgan declined to comment on the matter.
These developments have raised concerns among financial institutions that increasingly rely on artificial intelligence for software development, research, and internal productivity tools.
Anthropic recently restored its newest AI models
The latest reports also come shortly after Anthropic reinstated access to its newest AI models.
On July 1, the company resumed public availability of Claude Fable 5 and Claude Mythos 5 after U.S. authorities lifted export restrictions that had temporarily suspended deployment in June.
Anthropic said the restoration followed discussions with U.S. regulators and the implementation of additional safety measures.
Among the new protections were enhanced classifiers designed to detect and block cybersecurity-related misuse, particularly attempts to bypass the models’ safeguards through jailbreak techniques.
The company stated that these improvements addressed concerns raised by government officials regarding potential misuse of advanced AI systems.
Anthropic expands cooperation with U.S. authorities
Alongside restoring its latest models, Anthropic announced broader cooperation with U.S. government agencies.
The collaboration includes:
- Expanded AI safety testing
- Model security evaluations
- Monitoring for misuse attempts
- Information sharing on jailbreak techniques
- Ongoing risk assessments
Anthropic also emphasized that previously reported jailbreak methods affected only limited scenarios rather than representing a universal bypass of its safety systems.
The company maintains that its AI models continue to include multiple layers of security designed to prevent harmful use.
AI security remains a growing priority
The reported Alibaba decision reflects a broader trend among global enterprises as artificial intelligence becomes deeply integrated into software development and business operations.
Organizations are placing increasing emphasis on:
- Data security
- Model reliability
- Regulatory compliance
- Licensing restrictions
- Cybersecurity protections
For multinational companies handling sensitive intellectual property and customer information, AI governance has become as important as AI capability.
As governments introduce new regulations and businesses expand their use of generative AI, security reviews are likely to become a standard part of enterprise AI adoption.
Outlook
Although Alibaba has not officially confirmed the reported restriction, the news underscores the heightened scrutiny facing AI providers as enterprise adoption accelerates.
Anthropic has recently strengthened its safety measures and resumed deployment of its latest models, but questions surrounding licensing policies, regional availability, and security continue to influence corporate decisions.
As competition among AI developers intensifies, earning the trust of large enterprises will depend not only on model performance but also on transparency, security, and regulatory compliance.
