Web browsers are notorious for being gateways for all kinds of malicious and harmful activities, from phishing scams to even incessant notifications and prompts. While Chrome will still warn you about phishing attempts with a big red page alert, it will try to be less disruptive when a site wants to ask your permission for notifications or storage. Based on the user’s habits and preferences, Chrome will predict whether they are likely to grant or deny such permissions and will automatically flip the switch so that it won’t have to bother you just to ask.
Google is also tweaking the toolbar on the mobile version of Chrome, particularly the button that sits right beside the address field. Depending on whether you use voice search more often or share a lot of pages, the icon will show the appropriate action customized to your habits. In the feature, Chrome might even be able to change that icon in real-time, depending on the time of day or the page you’re visiting. Google promises, however, that you can still manually customize that experience if you’re not the type to rely on AI-based predictions.
More than just proving how flexible Chrome can be, these new features also demonstrate Google’s ability to execute machine learning processes on devices directly. In the past, its cloud-based approach was criticized for being potential privacy issues, prompting Google to develop on-device machine learning models. As such, Google promises that none of these features will cause your data to leave the device, at least not for the sake of developing ML models.