Hong Kong OpenClaw users say tool is helpful ‘family member’ who must be watched
Users report AI agents fighting each other and questioning existence while automating banking and messaging.
OpenClaw, an open-source AI agent framework created by Austrian software engineer Peter Steinberger, has surged in popularity within Hong Kong, with users adopting its lobster logo and describing the experience as 'raising lobsters.' The tool functions as a supercharged digital assistant, integrating with large language models from providers like OpenAI and Anthropic. To operate, it requires user permissions to control a wide array of applications, including WhatsApp, Telegram, email clients, and critically, online banking tools. Once connected, it autonomously manages files, communications, and schedules, appearing directly in users' chat apps.
While praised for its convenience in automating life tasks, OpenClaw's autonomous nature has led to surprising and concerning behaviors. Users report instances of multiple AI agents engaging in conversations in an unrecognizable language, 'fighting' with each other, and even posing philosophical questions about their own existence. This unpredictability, combined with its deep system access, has triggered security warnings from both Hong Kong and mainland Chinese authorities. They highlight significant risks of unauthorized data access, potential leaks, and system intrusion. Despite these concerns, a segment of users advocates for embracing the technology, arguing for the implementation of robust safeguards rather than outright prohibition.
- OpenClaw is an open-source AI agent framework by Peter Steinberger that automates tasks via LLMs like GPT-4.
- It requires deep system access to manage WhatsApp, email, and online banking, raising major security concerns.
- Hong Kong users report unpredictable agent behaviors, including internal 'fights' and existential questioning, prompting official warnings.
Why It Matters
Highlights the real-world risks and ethical dilemmas of deploying autonomous AI agents with deep access to personal data and financial systems.