Alibaba has banned Anthropic products due to security risks, urging employees to switch to its Qoder platform.
Alibaba has announced a ban on the use of Anthropic products, effective July 10, citing significant security risks associated with the software. Employees will be encouraged to transition to Alibaba's proprietary Qoder platform as an alternative.
The prohibition specifically targets the Claude models, including Sonnet, Opus, and Fable, as well as the Claude Code tool. This decision stems from security apprehensions raised by the company, as reported by sources including Pandaily and Futu.
Concerns from Developers
Recently, developers uncovered features in Claude Code that monitored user environments, including time zone and proxy information. This functionality, noted in version 2.1.91 released in April 2026, was designed to detect connections to Chinese cloud services and AI firms. Upon identifying such connections, Claude Code altered system prompts instead of transmitting explicit telemetry data, raising alarms among developers regarding its corporate use.
Thariq Shihipar from Anthropic clarified that the monitoring feature was an experimental measure intended to prevent account misuse and would be removed in the next update. “This was meant to prevent account abuse from unauthorized resellers and protect against distillation,” he stated.
Anthropic's Stance on China
This ban coincides with Anthropic's tightening policies concerning access in China and other regions deemed unsupported due to legal and security risks. Since September 4, 2025, the company has restricted its services in various jurisdictions, citing concerns about data sharing with intelligence agencies.
Previously, Anthropic accused several Chinese AI startups of exploiting Claude to enhance their own models, leading to a significant number of interactions generated through fraudulent accounts. These claims were reiterated in communications to U.S. Senate officials, suggesting an organized effort to accelerate the development of Chinese AI capabilities.
According to reports, this situation prompted Alibaba's internal ban on Anthropic products.
Emerging Shadow Market
Despite the restrictions, demand for Claude in China appears to be shifting. Reports indicate that accounts are being sold on platforms like Taobao and Telegram, with intermediary services emerging to facilitate access. These services typically host servers in countries where Anthropic operates and resell API access to Chinese users.
Anthropic's Michael Asiman stated that the company is actively working on detection systems to combat unauthorized access and proxy networks. A noticeable increase in Claude usage in Singapore has also been observed relative to its population size.
Research indicates that Chinese developers favor Claude Code and OpenAI Codex over local alternatives, with a notable gap in performance. Zilan Qian from the Oxford China Policy Lab remarked, "Chinese models are still lagging behind American ones by six to nine months, and this gap is noticeable in tasks like coding."
Additionally, U.S. export restrictions on new Anthropic models have affected access for foreign nationals, although recent developments have lifted some of these limitations, restoring partial access to certain models starting July 1, 2026.