Google DeepMindLaunchAI AgentsNovember 20, 2026

Google previews semantic governance policies for AI agents

Read original source →developers.googleblog.com

Available in Preview in the Gemini platform, the policies route every proposed tool call through Agent Gateway, where an LLM judge checks it against the user's request and plain-language rules before execution. Google says permissions alone can't stop indirect prompt injection or unauthorized data sharing.

1 source

More stories today

Open the live feed