Hong Kong government employees have been warned not to install the fast-growing artificial intelligence software OpenClaw on work devices after officials raised concerns about potential cybersecurity vulnerabilities and data exposure.

Internal notices circulated across public departments advised staff to avoid downloading or running the open-source AI agent on government computers and mobile devices connected to official networks.

Officials said the precaution was necessary because the software can require extensive system permissions and access to files and applications, potentially creating security risks if used on sensitive systems.

OpenClaw has attracted widespread global attention in recent months due to its ability to operate as an autonomous digital assistant capable of performing complex tasks on behalf of users.

Unlike conventional chatbots, the software can interact directly with computer systems—reading files, executing commands and connecting with other digital services to automate work processes.

Security specialists say these capabilities can make the technology powerful but also potentially dangerous if it is installed without proper safeguards.

Because the agent often requires broad access to a computer’s operating system and external platforms, misconfigured deployments could expose systems to cyberattacks, data leaks or malicious instructions delivered through third-party extensions.

The warning reflects a broader pattern of government caution surrounding the technology.

Regulators and cybersecurity bodies in several jurisdictions have issued alerts about OpenClaw deployments, noting that improper installation or unrestricted permissions could create vulnerabilities in corporate or public-sector environments.

Despite these concerns, interest in the software has surged across the global technology community.

Developers and companies have been experimenting with OpenClaw to automate tasks ranging from managing email and scheduling meetings to writing software code and coordinating workflows.

The rapid spread of such “agentic” AI systems has sparked debate among security experts about how organizations should balance productivity gains with potential risks.

Officials in Hong Kong said the precautionary guidance is intended to protect government systems and ensure that sensitive information remains secure while authorities monitor developments in the emerging technology.

Employees were advised to consult internal technology teams before installing any experimental AI tools on official devices or networks.

Cybersecurity analysts say the episode highlights the growing challenge governments face as powerful AI tools become widely accessible.

While autonomous software agents promise significant efficiency gains, their deep integration with computer systems means that careful governance and security controls will likely be required before they are widely adopted within critical public infrastructure.