China Restricts OpenClaw in Banks and Agencies for Security Reasons

Chinese authorities have moved to restrict the use of OpenClaw applications, an open-source artificial intelligence agent, on office computers within state-owned enterprises and government agencies. The action came in recent days, with warnings issued to prevent installations for security reasons, according to sources familiar with the matter. This includes the country's largest banks, where employees have been instructed not to download the software on devices connected to the company network.

OpenClaw, created by Austrian developer Peter Steinberger and uploaded to GitHub in November 2025, gained enormous traction in China since early 2026. It enables autonomous tasks like managing emails, scheduling flights, and integrating external tools, attracting mass experimentation by businesses and consumers. However, this popularity came with warnings: local governments in hubs like Shenzhen promoted the software with subsidies, while Beijing is now curbing its use in sensitive sectors to avoid cybersecurity risks.

According to Bloomberg, agencies and state-owned companies, including banks, were notified to check and remove existing installations, with some cases extending the ban to personal cell phones on the corporate network. A source indicated that the order came from regulators, highlighting concerns over the leakage of sensitive data, as the agent operates independently and accesses local files. Reuters confirmed that while it is not a total ban, employees were warned of dangers and instructed to obtain prior approvals.

This restriction reflects broader tensions in AI adoption in China, where the national "AI plus" plan promotes technological integration, but national security prevails. The software, now with Steinberger hired by OpenAI, runs locally and connects to models like those from Anthropic, facilitating usage but amplifying vulnerabilities such as malicious prompt injections or operational errors that delete critical files.

In the private sector, adoption remains strong, with giants like Tencent and Alibaba offering integrated versions, such as QClaw on WeChat. But in state areas, the measure might boost secure domestic alternatives. Analysts predict that by the end of 2026, over 40% of Chinese companies will adjust AI policies for compliance, according to industry reports, while the Ministry of Industry plans further reviews of exposure to open-source tools.