Google’s Gemini Spark AI agent can now browse the web and run errands for you

Google has integrated Gemini Spark with Chrome, allowing the AI agent to navigate websites and complete more complex online tasks using a user’s logged in accounts and saved passwords.

The new Chrome auto browse feature allows Spark to handle errands such as researching flights, beginning the booking process and scheduling viewings for apartments a user has previously saved. The feature requires user permission before accessing connected accounts.

Google said Spark is designed to return control to the user before completing sensitive actions such as payments. The company also said it has introduced protections against threats including prompt injection, where malicious instructions embedded in a webpage attempt to manipulate an AI agent.

Chrome browsing capabilities are initially rolling out in the United States, with additional regions expected to receive access later.

Google is also making Gemini Spark available to Google AI Pro subscribers in more than 160 additional countries. The expansion marks a significant increase from Spark’s initial rollout, which began with trusted testers and Google AI Ultra subscribers in the United States.

Introduced at Google I/O in May, Gemini Spark is positioned as a personal AI agent that can perform tasks across Google products and external services rather than simply respond to questions. The system runs on Google’s Gemini 3.5 Flash model and is designed to ask for approval before taking major actions.

The Chrome integration gives Spark access to the browser context required to complete longer workflows that involve multiple websites, saved credentials and user accounts.