Chrome sends AI history search data to Google
Google is working on a new feature in Chrome that gives artificial intelligence control over a user's browsing history. Chrome users may then interact with the AI when they run searches in their browsing history.
Here are the highlights:
- Chrome users may soon search the browsing history using natural language.
- The browser stores contents of visited pages locally in encrypted form.
- The feature submits search data to Google.
Google notes that search includes "general page content" as well as page titles and URLs. The feature returns "improved results" and works from the address bar and the history page according to Google.
Since it uses AI, it supports using "everyday language to search", according to Google.
Word of warning: Google displays a warning to users who manage the feature in Chrome. Under "things to consider", Google writes that data is always sent to Google when the feature is used.
Data includes the "history search terms, page content of best matches, and generated model outputs". Furthermore, Google confirms that the data may be accessed by human reviewers "to improve the feature". Chrome saves the content of pages "in an encrypted form" on the device.
This does not mean that Chrome reviewers do not get access to page contents, as MSPoweruser author Rafly Gilang suggested. Google even states that reviewers gain access to "page content of best matches".
There is no word about anonymization of personal data. It is possible that Google does that, but there is no evidence on the current "History search, powered by AI" page in the Chrome settings.
In other words, Google may gain access to personal information. A very simple example is a personal website or social media profile.
Good news is that the feature can be turned off or on in Chrome. Whether it is enabled by default or not remains to be seen. It would probably go against GDPR in the EU, if it would be turned on automatically.
Closing Words
Running searches across URLs, page titles, and content is certainly a desirable feature, provided that data stays local and users have full control over the feature.
The current history feature may only return matching URLs or titles. Extensions such as Memex were created in the past to allow users to search page contents as well.
Google's work on the AI-powered history search feature is ongoing. It is possible that things will change in future releases or that the entire feature is scrapped before official release.
Would you use an AI-powered history search feature?
Chrome is a cancer.
Actually, Google as a company is a cancer.
I will stop using their services all soon
They all should be paying us money everyday for all that data mining, and now AI data mining. It grinds my gears to think all those companies earning billions of dollars selling our mined data and we don’t get any money in return.
Is this proven or just clickbait?
Because most people have the little “send cookie information to search engine” which is the reason why you get rich responses by Google search on the omnibox, if you think that doesn’t contact the Google search servers to get suggestions well, it’s silly to think they are not using that to ‘improve’ their search results.
So it is pretty much easy to use the whole search history for the same.
People should disconnect from the internet and the world in general if they think they live in a world where they are not being watched, even your neighbor probably has a camera pointing at you knowing everything you do. So expecting to use the internet, and then having some mythological creature called Privacy at your side, well, you have been eating too much junk food that it already messed up with your brain.
Use Chrome, especially with an account in any Google owned service, and you are calling for them to your use information to improve their services, everything is toggled on by default, so even if they had options to turn things off, they know most people won’t turn the off.
You also know these people don’t care about your data, you are just an egocentric person who thinks a multi billion company cares about ‘you’, they just need data, random data, they can’t point the data exactly at you, but they still can get data to improve ‘ads’ or something in your area, ISP, IP and all that.
Nobody has proven otherwise, but people make it seem like Gov and everyone are not having access to tons of data that could potentially target you, but a multibillion company, they just care to generate more money and they do that (google) by showing their clients how accurate their data is to sell you a product, not to ‘spy on you’.
This same system has been used for voice and every piece of technology. Just look at videogames, they make balance changes based on every headshot, winning rate of weapons, what’s more popular and unpopular, what skill is used more than other, every piece of data is registered and saved in their servers… can they track you with it? unlikely, it’s just a bunch of random data making a big chart, would you say they are ‘invading your privacy’? well you should, because that’s what they are, but that’s what has been used in every industry to improve technology, you like it or not.
This sounds like one of the Borg saying “Resistance is futile. You will be assimilated!”
I, for one, will take what actions I can to resist. I’m not interested in exchanging my privacy for some kind of perceived convenience. I at least want to retain as much personal agency as possible in deciding what and when I make that trade-off.
That’s a wild spin and a lot of whataboutism. And I thought I was drinking the cool-aid because it is 95°F outside.
Well, they won’t be sending mine as I will never use Chrome, aka the browser for sheep.
yea sure. now its AI hypetrain that google uses to get access to our private stuff, ofc only in users interest. if its recall or this, spyware is spyware.
Why would anyone assume otherwise?
Appreciate the heads up. Wouldn’t touch this for any reason and imagine most who work in regulated industries wouldn’t either.
People wonder why DLP / data exfiltration risks are becoming greater concerns as a threat vector than network edge / pen-testing risk.
Makes you wonder who the malicious actors really are…
Color me surprised, not.
Good news is that the feature can be turned off or on. Bad news is that no one can be sure that flipping the switch off actually does anything more than flip the switch so you get a false sense of privacy. Kinda when you turn everything off in windows settings and believe you have turned things off… Ever fake-thrown a ball repeatedly with a dog? Yeah, you’re the dog now.
Don’t use a windows version which has settings :)
Or… find the equivalent settings in the registry, or via control panel, or via cmd.
No thanks I know AI is all about gathering data about a user. That’s why I don’t plan to use AI and I think many will realize that AI is too intusive.
Maybe they should just restore from backup what they used to use for search to the year 2010. That would be a 1000% improvement at $0.00 in development costs and a massive increase in user satisfaction. And, also,replace Sundar Pichai and Liz Reid with anyone or anything – literally a potted plant would be better.
About the concept : we’ve got the history feature in all browsers and, unless history is kept session after session, infinitely so to say, history without AI should AI allow “everyday language to search” seems to me just fine. Personal note : 1- History together with caches and search history wiped on browser exit, 2- A toolbar button has been added to toggle history hence within the session as well… so ‘AI History Search Assistance’ and I are unlikely to be pals :)
About Google and its eternal conquest of knowing what users do, what users think, who they are together with the thrilling challenge of ketchup-AI (uneducated eaters pouring the red sauce over anything they get to eat), they just had to add AI surveillance in their arsenal, especially that it serves their aims. OK, users are warned, least they could do, and Meta’s latest confrontation with the EU may not be stranger to this precaution, but, again, will the feature be opt-in or opt-out, or opt-in within the EU and opt-out elsewhere? One thing is sure, these GAFAM follow two policies when it comes to users’ rights : tangent to the law (trials show they’ve happened to break the curve) and profit to inconceivable heights. Bypass those intruders, period.
@ Tom Hawack,
You said, quote: “History together with caches and search history wiped on browser exit”, unquote.
That doesn’t amount to much if the data is sent to Google in real time. Clearing it on exit might make you feel better, but it didn’t make any difference to what actually happened to it and where it now resides safe and secure in the cloud and no longer within your reach.
Using a system that doesn’t include the infernal AI is the only viable solution, or at least for the time. Windows 8.1 user here with Floorp browser installed.
@TelV, my comment about wiping history and caches concerned the very concept of Google new feature in Chrome that gives artificial intelligence control over a user’s browsing history, emphasizing on my disinterest for such a concept given i don’t keep history and that I even happen to disable it.
When it comes to what is sent to Google, for one i don’t use Google’s Chrome browser, for two access to Google servers is blocked system-wide for most with a few exceptions though, exceptions handled by uBlock Origin and blocked by default.
Exceptions concern the minimum required for YouTube (though I use the ‘Piped” front-end but just in case all Piped instances would fail), and for a specific extension (TWP Translate Web Pages which needs only access to [translate.googleapis.com]),
Otherwise, I do keep a door opened, yet locked, for other Google services in case relatives, visitors would wish/need to access Google services, and to open the door I call a VPN extension (Windscribe, with a preference for Zürich and Paris, anyway a location with the same time zone as that of the OS) which will take the relay of my system-wide blacklists and DNS handled by DNScrypt-proxy, enabling therefor access to Google servers, for the very short time they are needed, after what i disable Windscribe and recover automatically my DNSCrypt-proxy DNS resolvers and blacklists …
The story of my digital life, isn’t it? Partially anyway. Believe me, the armor is solid as a rock. :)
I don’t care, I only search common things. Thanks for the article! :]
it is called spyware
The good news for Chrome users is that Google already had all of their personal information anyway. This probably just makes it easier and more efficient for Google to slurp everything up into the borg.