Google's AI can now browse the web as you

Google has rolled out two major updates for Gemini Spark, its AI agent designed to automate everyday tasks.

According to 9to5Google, the biggest change is support for Chrome's built-in browser automation on desktop.

Instead of working through a remote browser running on Google's servers, Gemini Spark can now interact with websites directly in the user's own Chrome browser. That means it can access already signed-in accounts, saved passwords, and other browsing data to complete tasks on the user's behalf.

Once permission is granted, Chrome displays a Gemini indicator in the toolbar so users can see when the AI agent is controlling the browser. Google says sensitive actions, including payments, still require manual approval. Chrome also includes built-in protections against prompt injection attacks, which attempt to trick AI agents into following malicious instructions embedded in webpages.

Google says Gemini Spark can handle repetitive web tasks such as scheduling apartment viewings from saved listings or finding flights and starting the booking process.

The Chrome integration is rolling out first in the United States, with other regions to follow. Google is also expanding Gemini Spark to Google AI Pro subscribers in more than 160 countries, after completing its U.S. rollout earlier this month.