What happened
Google added a new security layer in Chrome to support Gemini’s agentic browsing features. The update limits what AI-driven actions can run in the browser and adds guardrails to prevent unsafe or unauthorized behavior.
Who is affected
Chrome users who enable Gemini agentic browsing features, along with organizations that allow AI-assisted workflows in the browser.
Why CISOs should care
Agentic browsing introduces automation that can trigger actions inside the browser. This expands the attack surface. Without proper controls, AI-driven tasks can interact with sensitive data, execute unintended actions, or be exploited by malicious prompts.
3 practical actions
-
Review and update browser security policies before enabling agentic browsing features.
-
Enforce least-privilege access for browser extensions and AI tools tied to employee workflows.
-
Monitor for unusual automated browser activity that may signal misuse or prompt-based manipulation.