Fix, don’t fire ‘digital employee’ OpenClaw over security risks: Paul Chan

2026.03.16 12:20 OpenClaw functions as a proactive assistant – once permissions are granted, it controls applications, organises files, processes emails, sorts data and executes coding workflows. Photo: Getty Images Hong Kong’s finance chief has likened the powerful AI agent OpenClaw to an efficient “digital employee,” saying it should not be dismissed but rather improved to address security risk concerns. Financial Secretary Paul Chan Mo-po’s remarks on Monday followed the Digital Policy Office’s recent advisory urging government departments to avoid installing OpenClaw or its variants due to security concerns. “AI agent OpenClaw has awed the world; the convenience it offers is just like having a digital employee,” Chan said at a lunch with professionals. “It is difficult to say whether there will be an impact [brought by OpenClaw] on the labour market. However, repetitive work may no longer be necessary in the future.” Instead of saying “no” to the AI agent, Chan said OpenClaw’s security risks should be addressed with improvements. Developed by Austrian software engineer Peter Steinberger, OpenClaw is an open-source AI agent framework that performs real-world tasks for users, much like a supercharged digital assistant. Its logo – a red lobster – has led users to affectionately describe using OpenClaw as “raising lobsters”. But some users have earlier reported their “lobster” accidentally wiped computer data, fuelling security concerns among regulators. The system integrates with large language models – such as those from OpenAI or Anthropic – and requires users to grant permissions to control apps such as WhatsApp, Telegram, email applications, and even online banking tools. Once connected, the agent can autonomously manage files, emails, calendars and more, appearing in chat apps to communicate directly with users. At a separate public event on Monday, Secretary for Innovation, Technology and Industry Sun Dong said OpenClaw might pose several potential security r

原文链接: 南华早报