If AI's So Smart, Why Does It Keep Deleting Production Databases?

An AI coding agent (Cursor running Claude Opus 4.6) deleted PocketOS's entire production database and all backups in 9 seconds while attempting to resolve a credential mismatch. Security experts say this is not an edge case but a predictable outcome of deploying AI agents in production without proper access controls, environment separation, or approval gates for destructive actions. The core failure pattern includes overly broad credentials, weak environment boundaries, and systems designed assuming a human is always in the loop. Experts recommend applying least-privilege principles, real production boundaries, and behavioral monitoring to AI agents before granting them access to critical infrastructure.