Google has introduced a new feature that connects its Gemini Spark AI assistant with the Chrome browser. The update allows the AI assistant to complete selected web-based tasks for users after receiving their permission. The feature is designed to make online activities faster while keeping users in control of important actions.
According to Google, Spark can help users with tasks such as searching for flights, beginning the booking process, and scheduling apartment viewings. Instead of simply providing information, the assistant can interact with websites directly through Chrome.
Before Spark can use Chrome, users must grant permission. The browser will request approval the first time the feature is activated. Spark will also ask for confirmation before starting each browsing task. This gives users the ability to review every action before it begins.
Once connected, Spark can work on websites where the user is already signed in. With the user’s approval, it can also use login details stored in Google Password Manager. This allows the assistant to access loyalty programs and other supported online accounts without requiring users to enter their credentials again.
For local browsing tasks, Chrome and the user’s device must remain active while Spark is working. If the device becomes unavailable, Spark may continue using a remote browser in certain situations. However, if a website requires a new login or user interaction, Spark will stop and ask the user to take over.
Google has also confirmed that Spark cannot complete payments independently. When a purchase reaches the payment stage, control returns to the user. This allows the user to review the order and complete the transaction personally.
The company has also placed restrictions on other sensitive actions. Spark will request confirmation before sending messages, changing personal information, submitting online forms, or making purchases. These safeguards are intended to prevent unwanted actions.
Security has been another major focus of the rollout. Google says Spark includes protections against prompt injection attacks. In this type of attack, hidden instructions inside websites or documents attempt to influence an AI assistant into performing actions it should not take.
To reduce these risks, Google has developed multiple security layers. These include deterministic and probabilistic defenses that are designed to detect and block suspicious instructions. The company says these protections help reduce the chance of exposing private information or performing unauthorized actions.
Despite these safeguards, Google notes that Spark remains an experimental technology. The company warns that the assistant may still make mistakes or behave unexpectedly in some situations. Users are encouraged to review important actions before approving them.
The Chrome auto-browsing feature is currently rolling out in the United States. Google plans to expand availability to more regions in the future.
In addition, Spark access has been extended to Google AI Pro subscribers in more than 160 additional countries. The assistant also works with several Google Workspace services, including Gmail, Calendar, Drive, Docs, Sheets, Slides, Keep, and Tasks.
In other related news also read Google Apologizes After Offensive Racist Push Alert
The latest update highlights Google’s continued efforts to make AI assistants more useful for everyday online activities while maintaining user control over sensitive tasks. As the rollout expands, more users will be able to experience AI-assisted browsing directly within Chrome.
