【社評雙語道】善用AI智能體 保持清醒勿跟風

原文

摘錄自3月12日香港《文匯報》：近期，一款名為OpenClaw、圖標為龍蝦的開源AI智能體在內地迅速走紅，被戲稱為「養龍蝦」。從杭州蕭山到合肥高新區，多地政府紛紛推出補貼政策，鼓勵企業和個人部署這一新技術；然而與此同時，有消息指，國有企業、政府機關乃至大型銀行已收到通知，要求在辦公網絡環境下限制甚至禁止使用OpenClaw，以防範潛在的資訊安全風險。

這一「冰火兩重天」的現象，折射出當前人工智能（AI）技術發展中機遇與挑戰並存的複雜局面。對於以建設國際創新科技中心為目標的香港而言，這既是一次重要的啟示，也是一道必須審慎應對的考題。

OpenClaw之所以能迅速風靡，在於其降低了AI智能體的應用門檻。用戶無須深厚的編程基礎，即可通過本地部署的方式，讓AI助手處理郵件管理、日曆安排甚至代碼開發等任務。這種「僱用AI機器人幹活」的模式，極大提升了工作效率，激發了市場熱情。地方政府積極推動，正是看中了其在賦能產業升級、提升數字經濟活力方面的巨大潛力。

然而，技術的雙刃劍效應在此刻顯露無遺。工信部網絡安全威脅和漏洞信息共享平台明確指出，OpenClaw部分實例在默認配置下存在嚴重安全缺陷，極易誘發網絡攻擊與資訊洩露。

更令人擔憂的是，由於安裝調配具有一定門檻，市場上湧現出大量水平參差不齊的「速成師傅」，他們在缺乏有效監管的情況下提供代安裝服務，無異於將數字家門的鑰匙隨意交給陌生人，埋下了巨大的安全隱患。

新技術從來不是洪水猛獸，但盲目跟風、忽視風險則可能釀成大禍。在金融、政務、國防等關鍵領域，數據安全事關國家命脈，任何潛在的漏洞都可能被惡意利用，造成不可估量的損失。因此，在推廣應用的同時築牢安全防線，是確保AI技術健康發展的必由之路。

對於香港而言，擁有「內聯外通」的獨特優勢，能第一時間接觸全球前沿技術。面對OpenClaw這類新興AI技術，香港社會固然要認真看待，加強學習認識，鼓勵科研機構和企業積極探索其應用場景，發揮好橋樑作用。

然而，熱情之餘更需保持清醒。香港在金融、法律、醫療等關鍵領域高度依賴資訊系統的穩定與安全，若在這些領域盲目應用未經充分安全驗證的AI智能體，後果不堪設想。

因此，香港在推動AI技術發展時，必須堅持「創新與安全並重」的原則。一方面，政府應加快制定針對AI智能體的安全標準與合規指引，建立嚴格的審批與監測機制，確保技術應用在可控範圍內；另一方面，應鼓勵業界加強自主研發能力，提升對開源技術的安全加固水平，避免過度依賴外部未經審核的工具。

同時，可考慮設立專項基金，支持安全可信的AI解決方案研發，引導市場走向規範化、專業化。唯有如此，才能真正實現用好AI技術賦能高質量發展，讓科技成為推動社會進步的可靠力量，而非潛藏危機的隱患。

譯文

Leveraging AI for Development While Guarding Against Risks: Avoid Merely Following Trends

Recently, an open-source AI agent named OpenClaw, featuring a lobster icon, has rapidly gained popularity in the Mainland, earning the nickname "Lobster Farming". From Xiaoshan in Hangzhou to the High-tech Zone in Hefei, various local governments have rolled out subsidy schemes to encourage businesses and individuals to deploy this emerging technology. Simultaneously, however, reports suggest that state-owned enterprises, government departments, and even major banks have received notices to restrict or ban the use of OpenClaw within office networks to mitigate potential information security risks.

This stark contrast — enthusiasm on one side and caution on the other — reflects the complex landscape in which opportunities and challenges coexist in AI's ongoing development. For Hong Kong, which aspires to build itself into an international innovation and technology hub, this serves as both an important lesson and a challenge that must be handled prudently.

The reason OpenClaw has surged in popularity lies in its success in lowering the threshold for using AI agents. Users with little programming experience can deploy it locally and allow an AI assistant to carry out tasks such as managing emails, organising calendars, and even developing code. This model of "employing AI robots to work" significantly boosts efficiency and ignites market enthusiasm. Local governments have been keen to promote it precisely because of its enormous potential to upgrade industries and invigorate the digital economy.

However, the double-edged nature of technology is evident. MIIT (Ministry of Industry and Information Technology) Network Security Threat Information Sharing Platform has pointed out that, under default settings, some OpenClaw instances contain serious security flaws, making them highly susceptible to cyberattacks and data leaks. Even more concerning is the rise of numerous "quick-fix technicians" offering installation services without adequate regulation. Given the technical requirements involved, handing over installation to such unvetted providers is akin to giving strangers the keys to one's digital home — a major hidden security risk.

New technologies are never inherently dangerous, but blindly following trends while neglecting risks can lead to disaster. In critical sectors such as finance, public administration and national defence, data security is vital to national interests. Any potential vulnerability could be exploited maliciously, resulting in immeasurable losses.

Therefore, strengthening security protections while promoting adoption is essential for the healthy development of AI technologies.

Hong Kong enjoys the unique advantage of being closely connected with both the Mainland and the wider world, enabling it to access cutting‑edge global technologies swiftly. When faced with emerging AI tools such as OpenClaw, the Hong Kong community should of course study them seriously, deepen understanding, and encourage research institutions and enterprises to explore viable applications — playing well its role as a bridge. Yet such enthusiasm must be tempered with sobriety. Hong Kong's key sectors — finance, law, healthcare and others — rely heavily on the stability and security of information systems. Blindly adopting AI agents that have not undergone adequate security verification could lead to consequences too grave to contemplate.

Therefore, as Hong Kong advances AI development, it must uphold the principle of balancing innovation and security. On one hand, the government should expedite the formulation of security standards and compliance guidelines for AI agents, and establish robust approval and monitoring mechanisms to ensure that technology is applied within controllable bounds.

On the other hand, the industry should be encouraged to strengthen independent research and development, improve the security hardening of open‑source technologies, and avoid excessive reliance on unvetted external tools. At the same time, consideration could be given to setting up dedicated funds to support the development of safe and trustworthy AI solutions, guiding the market towards a more regulated and professional direction. Only by doing so can Hong Kong truly harness AI technologies to empower high‑quality development, allowing innovation to become a reliable driving force for social progress — rather than a source of hidden risk.

●Tiffany