Google may be preparing to build further on its efforts toward a more native nature for the web and more useful site features using Assistant in Chrome, based on a commit and demo spotted in the Chromium Gerrit code review. That all has since been pulled but reportedly centered around adding the AI assistant and a script specifically designed to help users interact with a given page or site to Google's Chrome for Android.
None of that means users will be talking to their browser in the same way they might in Google Maps, Assistant-enabled hardware, or the Google App. Rather than linking voice search features into Chrome, which would arguably be pointless, Google Assistant integration is all about cross-service data sharing and making services online more useful for users.
As the feature's title in the commits — Autofill Assistant — suggests, that means that certain information held by Google home will be usable across the web where appropriate. Google's own example points to one possible use as a way to speed up and improve interactions with rental or similar service locations based on what Assistant already knows.
For instance, Google Assistant will know if a user has booked a flight or other travel to a given location and might suggest, via a dedicated card, autofill details for a car rental when the user visits a business associated with that. It may also be able to help the user adjust the reservation later via Google Assistant itself without the need to return to the site.
What does this mean for privacy?
The features would depend largely on which sites choose to integrate Assistant and how those sites choose to utilize its data. Other examples might include its use to provide suggestions for hotel rental reservation dates or show concerts or events that will be in a given area the user will be traveling to during that timeframe.
Based on Google's recent push toward user-driven control over how their data is used, stored, and other privacy-based account settings, sites won't necessarily have access to that data at all until the user agrees to use what's shown on the card.
There may also be some settings that allow a user to change or revoke access on a site-by-site basis, as has been seen with the search giant's recent work to ensure that certain sensor data can be controlled at a more granular level. That's all been part of a larger push by Google, presumably to avoid further international fines and fees as well as to avoid some of the controversies that have riddled sites like Facebook — and itself, to a certain degree — over the past couple of years. So it's difficult to imagine that Google wouldn't approach the new features with due consideration for user privacy and security.
Coming to you …soon enough?
Another detail noticed in the now removed commits was the repeated inclusion of allusions to Google's upcoming I/O Developers Conference event. That's set to kick off on May 7 and runs through May 9.
That doesn't mean users will be able to start accessing the features in May though since the clues more directly point to demos at the event rather than a completely finished feature. That's indicated by the fact that surface details seem much more finalized than the features themselves and code listed with a "triggerGoogleIOStuff" designator. So it will probably be at least a couple of months before users on mobile are able to take advantage of the changes.