Google Cloud has unveiled AI Threat Defense, a new security platform that autonomously discovers code vulnerabilities, writes fixes, and deploys patches. The system leverages Wiz technology for IT environment mapping and attack vector prioritization, while a fleet of AI models handles monitoring — lightweight models run continuous network analysis and heavier ones activate only for critical node inspection.

When a vulnerability is found, Codemender, a DeepMind-built AI agent, replaces the problematic code directly in the development environment and can even refactor legacy architectures into modern programming languages. Before any patch goes live, the platform automatically generates tests to verify the changes.

Introducing Google AI Threat Defense