Advertisement
Artificial intelligence
Opinion
SCMP Editorial

EditorialChina must manage the risks of AI agents without stifling innovation

OpenClaw’s popularity shows authorities are right to urge caution, even as companies race to localise the new tech

2-MIN READ2-MIN
Listen
A person’s mobile phone runs the open-source AI agent OpenClaw in Huzhou, Zhejiang province, in this photo taken on March 11. Photo: Xinhua
China may be having its own TikTok moment. Just as Americans went crazy over the Chinese-made social media platform, mainland Chinese are in a frenzy to use OpenClaw, an American-owned artificial intelligence (AI) agent that promises to make the new tech accessible to the masses. And just like Washington, Beijing is worried about potential security risks and privacy breaches. Yet the business and tech potential of AI agents is too great to ignore.

On one hand, many local governments recognise opportunities that dovetail with national development goals. On the other, the central government sees risks that need to be managed. These are really two sides of the same coin and require careful management.

OpenClaw was acquired by market-dominant OpenAI in February. The AI agency programme promises to perform tasks and offer solutions for people. In China, unsurprisingly, many end up using it to pick stocks and advise on feng shui. Tech firms such as Hong Kong-listed MiniMax and Zhipu that offer OpenClaw tools saw their share prices rise. Stockbrokers are under pressure to adopt AI capabilities for clients. In major cities, people are seen queuing outside tech companies to install OpenClaw.
Advertisement

Chinese cybersecurity giant 360 Security Technology warns that the country is home to 40 per cent of global assets linked to OpenClaw. By any measure, that is an unusually high tech and financial exposure for any country.

So just as quickly, many are uninstalling it after China’s cybersecurity agency issued no fewer than two nationwide warnings about security and data risks tied to OpenClaw. Mainland and Hong Kong governments have warned civil servants against its use after some mainland local government departments and tech companies adopted the AI agent amid a nationwide frenzy.
Advertisement

The problem is that OpenClaw can access an unusual amount of personal data from people’s electronic devices. However, it may be too late for many to uninstall it. Some experts warn leaks may have already occurred and uninstalling it may not guarantee a clean break. Leading Chinese tech firms such as Alibaba (owner of the SCMP) – which launched the more secure JVS Claw – are rushing out competitive models. Together, they can form an ecosystem to localise the new tech and neutralise foreign threats. As Chinese commercial entities, they can work closely with regulators to better protect users.

Advertisement
Select Voice
Select Speed
1.00x