Alibaba has instructed its employees to remove all Anthropic products from their corporate devices after an internal audit identified possible backdoor risks in the AI programming assistant Claude Code. The measure, which takes effect on July 10, marks a relevant shift in the Chinese giant's strategy regarding the use of external tools.
The decision also includes a ban on Anthropic's entire line of models, such as Sonnet, Opus and Fable. The restriction halts the use of one of the solutions most widely adopted by the company's engineering teams and reinforces a broader move toward control over third-party technologies.
Until recently, the company encouraged the use of public AI tools. Since the beginning of 2026, engineers could receive monthly reimbursements of up to US$ 1.400 to use solutions such as Claude Code, GPT and Gemini. The program was widely adopted, with developers spending hundreds of dollars per week before requesting compensation.
Under the new policy, that incentive is ending. Alibaba will no longer subsidize external tools and is directing its efforts toward its own solutions, such as its internally developed Qwen family of models.
The restriction comes amid rising tensions between Anthropic and Chinese users. In communication with United States authorities, the company said it had identified more than 28,8 million interactions with its system carried out by about 25 thousand accounts considered fraudulent over a period of a few weeks. According to the company, the activity was intended to accelerate the development of competing technologies through so-called model distillation.
In addition, independent analyses indicated that recent versions of Claude Code contained functions capable of checking local settings, such as time zone and API or proxy parameters. Although developers classified some of these functions as experimental, the suspicions gained relevance after concerns were raised internally by Alibaba.
Despite this, there is no public confirmation that an intentional backdoor exists. The lack of official technical details keeps the debate open among experts.
Alibaba's move comes in parallel with similar restrictions adopted by global financial institutions, in addition to limitations imposed by Anthropic itself on international access to its more advanced systems.
With the new directive, the Chinese company reinforces its dependence on domestic AI infrastructure and expands the distancing between technological ecosystems at a time of growing fragmentation in the sector.