OpenClaw AI agent craze sweeps China as authorities seek to clamp down amid security fears — adoption surges as state-run enterprises are barred from use

If the big AI story in early 2025 was data centers, in early 2026 it's OpenClaw and AI agents — and somehow, it's made an even bigger impact in China than in America.

As Bloomberg reports, Chinese companies, individuals, and government entities have raced to adopt the open-source agentic AI tool so fast that it's reached a near cult-like status. There are hats, install parties, cloud-based options, and copycats appearing on smartphones. Indeed, it's gotten so big so fast that Chinese authorities have announced a major crackdown over security concerns and a growing loss of control over the new AI phenomenon.

OpenClaw makes using and integrating AI agents with existing software platforms much easier by working through a messaging interface. It's free, open-source, and has garnered heaps of interest in recent months for heralding the agentic AI era — so much so that it's leading to shortages of high-end Macs to run it. OpenAI hired its creator in mid-February, and Nvidia is concerned enough that the hardware giant is developing its own version of the tool for enterprises. OpenClaw is now one of the fastest-growing repos on GitHub, overtaking React, Python, and Linux in the starred list.

But even that frenzied excitement is nothing compared to adoption in China.

Chasing (lobster) tails

When China's DeepSeek dropped its AI chatbot in January 2025, the world was caught by surprise. Suddenly, the enormous software lead the U.S. Has in this burgeoning field was seemingly stripped away. Since then, China hasn't had a similar headline-grabbing moment. It's battled the U.S. Over access to important raw materials, secured some access to Nvidia GPUs, and put a rocket on the back of its domestic chip industry. But while Chinese AI developers are still putting out impressive new models, they're not grabbing industry attention or creating another DeepSeek moment.

That may well be what China is pursuing with its rapid adoption of OpenClaw. According to Bloomberg, there are gatherings of OpenClaw enthusiasts who wear specialized lobster hats at meetups, get-togethers of hundreds of people at Tencent offices to get help setting up the agentic AI tool, and there's even a growing vocabulary building around "Raising the lobster" — a colloquial way of discussing installing the software.

Companies have responded to the explosion of interest, with cloud providers offering virtual machines to run OpenClaw, saving interested citizens from setting it up themselves locally, and removing one more barrier to entry. Tencent, Alibaba, and AI disruptors like Moonshot and MiniMax are all offering their own tweaked versions of OpenClaw, and their stock prices have ballooned along with the hype. Minimax is now valued at $44 billion, despite only making $79 million in revenue in 2025.

Some local governments are even offering subsidies worth hundreds of thousands of dollars to companies with approved OpenClaw projects, so keen are they to foster adoption in their districts.

But OpenClaw is not commercial software. It's not an established platform with years of development behind it. It has no public face you can complain to when something goes wrong. It's a new platform that offers impressive potential, at enormous costs to security and privacy.

One implementation deleted the Meta AI head of alignment's entire inbox when it went rogue and ignored her commands. AI agents are exposing API keys, passwords, and private information, and making it easy for scammers to steal data and cryptocurrency.

It's for these reasons, and the runaway nature of the technology, that Chinese authorities are starting to crack down on it.

And then came the crackdown

Chinese authorities have issued notices to state-run enterprises and government agencies not to install OpenClaw on office systems, and that any existing installations must be declared for security checks and potential removal, Bloomberg reported. This crackdown particularly focuses on state-run banks, where employees are also being told not to install OpenClaw on their personal devices, and to uninstall it if they already have.

Other companies have received similar notices, though not always outright banning the agentic AI tool, but instead demanding prior approval from local authorities before proceeding.

OpenClaw can leave gaping holes in a device's security, which in turn can open up entire organizations to theft, infiltration, and sabotage. This poses grave concerns for Chinese authorities, who face the cult-like nature of OpenClaw's rapid adoption as opening the country up to dangerous security threats from domestic and international actors.

As Bloomberg suggests, there's also concern among the Chinese ruling party about losing control of this powerful technology. It previously cracked down on the power of internet firms in the early 2020s with new anti-monopoly laws when some of the largest held what it perceived as too much centralized power.

There's no suggestion the government will crack down on individuals or entrepreneurs using OpenClaw for now, but spokespeople have been conducting interviews with press that urge caution. That may be a sign of further restrictions to come, if the technology continues to run rampant — and perhaps even amok. As the AI industry grips for a wave of Agentic products, security and privacy become an ever-prescient topic, especially as users cede control to our new claw-wielding assistants.