China's AI Safety Split: Beijing Warns, Shenzhen Pushes Ahead

China's central government warned state agencies in March against using OpenClaw, an AI tool that can access passwords and accounts, citing security risks. At the same time, Shenzhen offered government workers the same tool with subsidies to encourage use. The contradiction shows how China's regions handle new AI differently — and raises real questions about whether giving autonomous systems access to sensitive credentials is safe.