Home AI Gemini Spark Can Now Browse the Web Inside Chrome

Gemini Spark Can Now Browse the Web Inside Chrome

Local Chrome auto browse for agentic web errands

6
0
Gemini Spark integrated with Chrome browser interface
Image: Google

Gemini Spark can now drive Chrome on your computer. With permission, Google’s agent uses your logged-in sites and saved passwords to run multi-step web errands.

Moreover, this is a local Chrome handoff, not just a remote sandbox browser. Google says Spark can start tasks like apartment viewing schedules or flight research and booking flows while you stay signed in.

What Chrome auto browse unlocks

Previously, Spark could browse the web in other ways. Now it can connect to the Chrome browser on your own device. After you grant access, Chrome can show a Gemini and auto browse indicator in the top bar while the agent works.

Advertisement

In addition, Google’s examples stay practical and narrow. Spark can research flights and start a booking. It can also schedule viewings for apartments you already saved. Google has not published a long list of supported sites yet.

The agent still needs you for money. Google says Spark hands payment steps back for review instead of finishing a purchase on its own. That keeps a human in the loop for the riskiest clicks.

Security and control

However, handing an AI your passwords is the real trust test. Google says it is investing in layered defenses against prompt injection, including both deterministic and probabilistic checks. Those attacks try to hide bad instructions in web pages or user content so an agent takes unwanted actions.

Google points to its broader agentic security work for the threat model. Concrete tool details stay light in the July 30 product post. Treat the protection claims as vendor language until Google shows more technical write-ups.

Who gets it, and what else expanded

Still, the Chrome integration is rolling out first in the U.S., with more regions planned. Google also says Google AI Pro subscribers in over 160 additional countries can start using Spark for daily task automation.

Therefore, Spark is no longer only a U.S. AI Pro experiment. It is becoming a wider agent product that sits next to Workspace integrations and earlier U.S. Pro access. For prior context, see our notes on Spark expanding to all U.S. Google AI Pro subscribers and the original Gemini Spark agentic assistant launch.

Finally, agent browsers will only feel safe if users can see every step and reclaim control fast. Chrome auto browse is useful for tedious forms. It is also a reminder to review account permissions, password manager settings, and any agent logs Google surfaces in the Gemini app.