Google Gemini Spark gains Chrome browsing integration

Google has expanded its agentic AI assistant Spark with the ability to work inside the Chrome web browser. Once a user is logged into Chrome, Spark can rely on their logged-in accounts and saved passwords to handle routine web errands. Google describes these as tedious tasks, offering two examples: researching flight options and starting the booking process, as well as scheduling apartment viewings.

The integration raises privacy and security considerations, given that the assistant gains access to sensitive credentials. Google says it has implemented safeguards against prompt injection, a technique where malicious instructions hidden on websites or in user-generated content could trick the assistant into performing unwanted actions, such as initiating financial transactions or exfiltrating sensitive data. However, the company has not shared concrete details, stating only that it is investing in a layered defense combining deterministic and probabilistic methods to make attacks difficult and costly.

Another safeguard is that Spark will not complete payments automatically; it will hand the process back to the user for review and finalization. The new browsing feature is currently rolling out in the United States, with additional regions planned later. Separately, Google has just expanded Spark access to Google AI Pro subscribers in more than 160 countries. The assistant also integrates with Google Workspace apps. Engadget has reached out to Google for more specifics about how the integration will affect users and will update the story when a response arrives.