Amid the buzz surrounding OpenClaw, a paid service for removing the AI agent has emerged on Chinese social media. Previously, users paid to install the software, according to SCMP.
On the Alibaba advertising platform Xianyu, numerous listings have appeared for "remove OpenClaw" services. A user from Shanghai, known as mojito lime water, is offering to perform the task for 299 yuan ($43.55) and has already completed dozens of similar requests.
Similar services are being offered in other major cities.
The sudden shift in interest towards OpenClaw followed a warning issued by China's National Computer Incident Response Center. The agency stated that the software for launching AI agents has extremely weak default security settings.
The China Academy of Information and Communications Technology announced an initiative to develop standards for so-called Claw agents to address the issue of their opaque decision-making processes.
The institute plans to gather proposals from companies and industry experts to create requirements called Reliable Capability Requirements for Intelligent Assistant Agents (Claw) Products. These will include standards for quality control and reliability of AI agent behavior.
The Hype Around OpenClaw
Launched in 2025, OpenClaw quickly gained popularity for its ability to autonomously perform tasks for users. It can sort or respond to emails, write work reports, and prepare presentations.
The software became particularly popular in China, where people lined up and were willing to pay for the installation of AI agents on their devices.
According to Zhou Hongyi, co-founder and chairman of the Chinese cybersecurity company 360 Security Technology, over 40% of approximately 150,000 installations are in China.
"In the age of AI, the largest attacks may no longer stem from server vulnerabilities but from intelligent agents granted the authority to perform tasks," the expert stated.
Several universities in China have issued warnings regarding OpenClaw. They recommend using only the latest official version of the app, limiting the program's internet access, and closely monitoring permissions.
Users have also been cautioned against risky actions such as installing third-party mirror versions, visiting unverified websites, and disabling detailed log audit functions.
In February, Meta AI security researcher Summer Yue tasked the AI agent OpenClaw with checking her overflowing email. The bot began deleting everything at lightning speed.