Notes from the Asia-Pacific region: China's 'Two Sessions' looks to country's AI future

With a warm breeze, spring has arrived in China and the data and artificial intelligence landscape is heating up.

March is a pivotal month in China as it is when the "Two Sessions" of the National People's Congress normally convene in Beijing, with top leaders and stakeholders in various fields and sectors gathering to discuss and set China's development plans and priorities for the upcoming 12 months. 

In the Central Government Work Report delivered by China's Premier Li Qiang at the two sessions, AI is positioned as the driving force for new quality productive forces, shifting China's digital economy from data-driven to intelligence-driven. The government will actively promote the development and deployment of new-generation smart devices and agentic AI, while providing strong support for building open-source AI infrastructures. To balance innovation and risk, the government will continue enhancing AI governance. 

Notably, during the Two Sessions, China's Supreme Court announced an expedited legislative process for AI laws and regulations. Courts will make efforts to handle more AI-related cases, clarify the delicate boundary between innovation and infringement, and draft more judicial policies to guide businesses on difficult issues such as the originality of AI-generated content and legality of training data. 

In March 2026, the open-source AI agent project OpenClaw — nicknamed "Lobster" because of its icon — went viral in China and became a major tech phenomenon. Powered by autonomous decision-making, multitool collaboration and complex task automation, it spread widely among developers and even ordinary internet users. 

Meanwhile, significant security and privacy risks emerged due to OpenClaw's architecture, including plug-in vulnerabilities, prompt injection, data leakage and unauthorized access. Regulators in China have quickly issued risk alerts, requiring enterprises to restrict access to sensitive data, strengthen access control, audit logs, anonymize data and ban unprotected use in financial and other key industries. These alerts have started to play an important role in helping users understand and manage the risks in "raising the lobster," that is, using OpenClaw. 

A similar regulatory approach has been adopted in Hong Kong. The Office of the Privacy Commissioner for Personal Data issued a 16 March alert reminding business organizations and members of the public to pay attention to and understand the personal data privacy and security risks involved to avoid personal data breaches, malicious system takeovers and cybersecurity threats before deploying or using OpenClaw and other agentic AI. 

To minimize the risk, the PCPD offered practical suggestions, such as granting minimum access rights to agentic AI, using the latest official version, adopting adequate data and cybersecurity measures, installing and using plugins with caution and conducting continuous risk assessments.