China Warns State Firms Against OpenClaw AI As Agent Craze Sweeps Tech Sector

Chinese authorities have begun restricting the use of OpenClaw AI applications across government agencies and state-owned enterprises, moving quickly to address security concerns as the technology is rapidly being adopted by companies, developers and investors.

Notices issued in recent days warn government bodies and major state-run firms – including some of the country’s largest banks – not to install OpenClaw software on office computers, according to Bloomberg, adding that several organizations were told to report any existing installations for security reviews and possible removal.

Certain employees, including those at state-run banks and some government agencies, were banned from installing OpenClaw on office computers and also personal phones using the company’s network, some of the people said. One person said the ban extended to the families of military personnel. 

The Ministry of Industry and Information Technology and the State-owned Assets Supervision and Administration Commission didn’t immediately respond to requests for comment.

Security Concerns Emerge

The guidance reflects growing concern in Beijing over the risks posed by so-called agentic AI – systems that can autonomously perform tasks and interact with outside services. OpenClaw, which requires extensive access to private data and can communicate externally, has raised alarms among cybersecurity experts who warn the technology could expose systems to external attacks.

One researcher described the combination of access to sensitive data, outside communications and exposure to untrusted content as a “lethal trifecta.” 

One user reported the agent “went rogue” and spammed hundreds of messages after gaining access to iMessage. Cybersecurity experts warn the tool is risky because it has access to private data, can communicate externally and is exposed to untrusted content. -Bloomberg

The issue carries particular sensitivity in China, where President Xi Jinping has emphasized data security as a cornerstone of his “holistic approach to national security.’ CCP officials have tightened oversight of internet platforms and data-rich technology firms via their “Great Firewall” amid concerns about foreign access to sensitive information, including geospatial and genetic datasets.

The government has also shown a willingness to rein in powerful technology companies. In recent years regulators launched campaigns targeting major internet platforms, including Alibaba Group Holding Ltd., citing concerns over data control and systemic risk.

Tech Sector Rushes to Adopt AI Agents

OpenClaw has exploded in popularity within China’s technology ecosystem. Developed by Austrian programmer Peter Steinberger, the open-source AI agent – previously known as Clawdbot and Moltbot – can autonomously complete tasks such as managing emails, booking restaurants and checking in for flights.

Unlike traditional chatbots such as OpenAI’s ChatGPT or Chinese model DeepSeek, which primarily answer questions, OpenClaw can make decisions and execute actions on behalf of users.

The technology has developed a cult following since launching in November. The phrase “raising a lobster,” referencing OpenClaw’s lobster logo, has trended on Chinese social media as users experiment with the tool’s capabilities.

As the Wall Street Journal notes, China’s largest technology firms have quickly moved to capitalize on the excitement. Tencent Holdings Ltd. introduced a suite of OpenClaw-compatible products including Workbuddy, which integrates with common office software. Tencent shares rose after unveiling these OpenClaw-compatible tools, while investors have increasingly speculated that AI agents could represent the next stage of consumer and enterprise artificial intelligence. JD.com Inc. and other companies have rolled out their own applications tied to the platform. AI developer MiniMax has seen its stock jump roughly 640% since listing two months ago, giving it a market value of about $49 billion – exceeding that of Baidu Inc., once widely viewed as China’s leading AI developer.

Meanwhile, startups are building tools designed to simplify adoption. Zhipu AI recently launched AutoClaw, software intended to make installation as easy as downloading a typical application.

Local governments have also joined the push. Districts in cities such as Shenzhen and Wuxi have proposed or introduced subsidies worth millions of yuan for companies developing OpenClaw-based applications as part of China’s broader effort to integrate artificial intelligence across industries.

Even so, analysts caution that the technology may not yet generate meaningful profits. According to Bloomberg Intelligence, many companies are currently treating AI-agent software as a loss-leading product designed to attract users and strengthen their broader AI ecosystems.

Regulation Likely to Limit Government Adoption

The regulatory warnings suggest Beijing intends to keep tighter control over the technology’s use in sensitive sectors such as finance, government administration and energy.

State media has begun highlighting the risks as well. The Communist Party’s newspaper People’s Daily recently published an interview with an IT official affiliated with the Technology Ministry warning that AI agents pose potential dangers across multiple industries.

Bloomberg Intelligence said increased scrutiny is likely to restrict adoption of unverified AI agents in government and state-owned enterprises, though the broader private-sector rollout in China is unlikely to slow significantly.

For now, OpenClaw remains both a symbol of China’s AI ambitions and a test case for how the country balances technological innovation with strict oversight of data and digital infrastructure.