# sandboxing

China Issues Security Red Flag on Open‑Source AI Agents as Domestic Firms Rush to Lock Them Down

China’s industry regulator has issued security guidance for OpenClaw, a popular open‑source AI agent framework, after monitoring showed many instances running with unsafe defaults. Domestic tech firms are racing to mitigate risks by offering cloud‑hosted, sandboxed and permissioned agent services, while legal and regulatory pressures—illustrated by a recent US court ruling against an autonomous agent—are starting to shape the market.