China moves to curb Openclaw AI use for banks, state agencies

China has begun implementing restrictions on the use of OpenClaw AI in financial institutions and state agencies, reflecting growing regulatory concerns over data security and privacy risks. The open-source AI agent, developed by Austrian engineer Peter Steinberger, has gained rapid adoption in China for its ability to automate tasks such as booking flights and managing emails, but its access to sensitive user data has drawn scrutiny.

Regulators have flagged potential vulnerabilities, particularly in sectors handling personal and financial information. For instance, the Doubao AI phone, which integrates an agent with system-level permissions to interact across apps, triggered widespread security alarms when major platforms like WeChat and Alipay blocked it over fears of unauthorized data access and fraud. These concerns align with Beijing’s broader emphasis on tightening data privacy rules since 2021, including stricter export controls and cross-border data transfer regulations.

While Shenzhen’s Longgang district has promoted OpenClaw through subsidies and incentives for “one-person companies,” it has also underscored the need for caution. Draft policies released by the district highlight the central government’s push for AI-driven industries but include provisions to mitigate risks, such as requiring localized data storage and limiting cloud transmission. Financial institutions are reportedly developing internal guidelines to govern AI agent integration, with some advocating for mandatory user authorization before sensitive transactions.

The debate underscores a broader tension between innovation and security in China’s tech ecosystem. As agentic AI challenges existing data governance frameworks, regulators face pressure to balance economic growth with safeguards against cyber threats and privacy breaches. The outcome of these efforts could shape global standards for AI regulation in critical sectors.