Google's Gemini Spark Gains Chrome Auto-Browse Control, Expands to 160+ Countries

Google is rolling out two updates to Gemini Spark, its personal task-automation agent: direct control of desktop Chrome for web browsing, and expanded international access for Google AI Pro subscribers.

Chrome auto-browse replaces remote browser

Gemini Spark previously relied on a remote web browser to complete tasks. Starting today, it can use the desktop Chrome browser installed on a user's own machine, according to 9to5Google. This allows Spark to act using accounts a user is already logged into and passwords saved in Chrome, rather than operating in an isolated remote environment.

The capability draws on Chrome's auto-browse feature, which Google introduced for Gemini in Chrome earlier in the year. Once a user grants access, a Gemini and auto-browse indicator appears in Chrome's top toolbar to signal when the agent is active.

Google cites examples like scheduling apartment viewings from saved listings or researching flight options and beginning the booking process — tasks the company describes as "tedious web errands."

On safety, Google says users must manually approve payments and other sensitive actions before Spark can complete them. The company also says the browser integration includes protections against prompt injection, a known attack vector where malicious web content attempts to hijack an AI agent's instructions.

The Chrome auto-browse feature is rolling out first in the US, with Google saying additional regions will follow. No specific timeline for international availability was disclosed.

AI Pro expands to 160+ countries

Separately, Gemini Spark is now available to Google AI Pro subscribers in more than 160 additional countries as of today. This follows a full US rollout completed earlier in July 2026.

Gemini Spark launched in May 2026 as Google's agent for automating personal and daily tasks. Since launch, Google has steadily added capabilities, including Connected Apps integration with Workspace and Search, a Personal Intelligence feature, a remote computer with code execution, and Canvas. More recent additions include MCP (Model Context Protocol) integrations and a Mac app enabling local desktop control.

Pricing for Google AI Pro and any Spark-specific costs were not detailed in Google's announcement.

What this means

The shift from a remote sandboxed browser to direct desktop Chrome control is a meaningful capability jump: it lets Spark act as the user, with access to real logged-in sessions rather than a blank slate. That's more useful for real-world errands like bookings, but it also raises the stakes on Google's stated safeguards — payment approval gates and prompt-injection defenses — since a failure mode now touches accounts with actual credentials and financial exposure, not a disposable sandbox.

The simultaneous international expansion signals Google is treating Spark as a core, scaling product rather than a US-only experiment, putting pressure on competitors offering browser-using agents to match both the capability and the geographic reach. Whether Chrome auto-browse proves reliable enough for unsupervised errands, or whether users limit it to low-stakes tasks, will determine if this becomes a genuine daily-use feature or stays a novelty confined to approved, sensitive-action checkpoints.