So I would have to suggest, use these features with extreme caution on any page you consider private if you aren't prepared for your private information to get sucked into Google's Gemini training data.
"How your data is used Google uses this data, as described in our Privacy Policy, to:
Provide our services Maintain and improve our services Develop new services Personalise our services (learn more) Customise our services Communicate with you Measure performance Protect Google, our users and the public
These uses extend to the generative AI models and other machine-learning technologies powering our services."
Does "Maintain and improve our services" mean "any private web page you ask Gemini about will be dumped into our training data"?
I've still not seen a solid answer to that from any of the AI labs that use language of that nature.
Safe to assume they will use your data for pretty much anything they can, including model training.
'When you use the Gemini in Chrome feature, Gemini collects and processes page content and the URL from the browser tab you’re viewing by default. Some of the page content Gemini uses might not be visible to you."
I think they're being reasonably explicit about what they're doing.
Note: I'm in the EU so not sure if this is what's shown everywhere else.
I'd love to get a confident answer to that question.
I never asked for it, or agreed to anything having to do with it. I'm pissed off to find it bound to a hotkey or gesture on my phone (I'm still not clear what the actual gesture is that keeps invoking the damn thing).
The more unsolicited crap Google jams down the pipe at me, the sooner they're going to discover I'm not at the other end and the pipe feeds straight into a septic field.
If anyone from Google is reading this, I hope you're ashamed of your dark patterns. You used to be a testament to the ideal of putting the user first. Now I can't distinguish you from any other crummy, misleading, self-serving tech gorilla.
Google is in the same position, yes they have Android, GCP, Gmail/work suite, etc but even all of that combined couldn't sustain the moloch.
These insane levels of revenue streams cannot be sustained with just one product for longer periods of time. Eventually, every one of these biggest X'es will need to branch out to different industries/domains to sustain those levels.
Of course they are not a 1:1 replacement, but for the first time we see Google's model being challenged and them having to defend, which they do by trying to drive competition of by integrating genai into their products. Once the competition is gone, they can reconsider.
Main difference is that Google is bad at monitising any (non advertising) service; so free becomes the main proposition.
Is Meta much different though?
It is for example a bit more than Tesla, double that of Broadcom or Oracle, a bit less than Dell or TSMC.
Seems they're actually pretty good at monetizing non-ads goods and services at this point.
Their advertising revenue is over $71B/quarter.
As far as I know the majority of the non advertising revenue is from Google Cloud and subscriptions for services.
Most of the interesting product initiatives they've launched (that I can think of) have folded.
But aside from that part, yes, I can't see how you could make a different assumption.
Your [ORGNAME] chats aren’t used to improve our models
So they definitely understand that people want to hear that their data isn't being used for training, and they know how to say it clearly and reassuringly. Which makes the omission of that in their consumer products more telling in my view.
But Google needs to leave this option open in the future, in case they have to go all-in on an arms-race against China, if Chinese AI starts becoming an actual threat somehow. And it's easy to predict the USA gov would prioritize that race over privacy concerns.
* Be 18 or over and in the US.
* Use a Mac or Windows computer.
* Use the latest version of Chrome.
* Have Chrome’s language set to English (United States).
* Sign(ed) in to Chrome.
Edit: reference https://abc.xyz/investor/founders-letters/ipo-letter/#:~:tex...
Except when you literally trick people into providing their data: https://x.com/dmitriid/status/1664682689591377923 by pretending that your dark patterns are "explicit consent"
Or except when you literally trick people into providing their data by connecting totally unrelated services https://x.com/dmitriid/status/1908951546869498085 and you don't stop until the person gives up and submits to you.
Except when your own support explicitly says that your behaviour is tracked across completely unrelated Google services: https://x.com/TeamYouTube/status/1849952594992435493
Except when you literally sign people into user accounts automatically with most data collection options turned on.
Except...
Can you provide a reliable source to verify it?
It's not nothing, but it's something. And, at least on my phone, it's not obvious if it can be turned off.
https://scholar.google.com/scholar?lr&ie=UTF-8&oe=UTF-8&q=Fe...
FUTO Keyboard is quite nice.
The integration with whisper is nice too.
https://support.microsoft.com/en-us/topic/microsoft-swiftkey...
Always assume companies will gather, use and share your data in all ways they legally can. The burden of proof is never on the user that companies don't milk us. Calling it "misinformation" as someone further above did is bizarre. This is the default business model of big tech.
Facts do matter, and I appreciate those that make an effort to state them correctly.
But to answer the question, Gboard absolutely uses your data. And it's right there in its privacy policy.
Hmm, no? It has access to all of the content of all of you're currently open tabs, and is able to parse images on web pages as well.
It would be neat if it could also browse on your behalf, but that would present all kinds of security risks.
But whats the vision of this? Where are they trying to take the customer?
I feel like this issue relates back to the origin of Google (search) in the first place. It was borne out of a technology in which the founders did not envision what it would become. It seems the firm just tries ideas and then tries to figure out where it goes - thats the culture. And unsurprisingly, yields a lot of failiures.
In contrast, Apples approach yields a much higher rate of success with less risk.
I do need machines to do the browsing for me.
They exist for that very reason, the web is much friendlier.
I 100% don't feel comfortable letting my browser work alone, but "agentic browsers" are a thing some people want and/or are building.
This is supposed to be a good feature? Not a privacy nightmare?
Of course, it should also be possible to completely disable Gemini so as to avoid accidentally sending it private browsing content.
Well, they're gonna have to support an astronomical scale of queries - not many companies in the world are able to do it and Alphabet is doing it pretty much on their own stack of cloud, a.i chips and software. So sure, the front end is not a big deal but this is still a big move.
https://research.google/blog/mechanism-design-for-large-lang...
Think TikTok, except where the platform is both curator and creator.
So just like how we don't watch the same thing at the same time anymore due to on-demand media, and the talking about yesterday's big TV show is only a thing of the past now. It will be one more step removed from that and you will have kids talking about this random thing that appeared in their custom show yesterday. Conversations like "dude, did you also get that singing toilet in your stream yesterday, what was that about".
Due to the laws of enshittification they will eventually never be tuned to your benefit.
Nothing wrong with that, in theory and in moderation.
I was 60% Chrome and 40% Firefox, now I'm 99% Firefox and 1% Chromium.
I built an extension like this with Claude-code a few days ago because I wanted to see if I could replace the ai feature of Firefox when I switched to LibreWolf. Turns out, it was quite easy for Claude code.
I want a bit further and tried to get the extension to browse around. Individual actions worked, but I couldn’t get it to follow a plan. In the end I finally looked around the code and Claude had made a huge mess with cursor etc.
The complexity of handling the array of messages was a bit too much for the AI agents.
I now have the same as this Gemini ai though and it CAN click links and it works with ollama too. So more private.
All in a few hours of development.
So I am not impressed by Google here
You can see which ones your Firefox installation has already fetched and their purpose on about:addons.
More about this here: https://support.mozilla.org/en-US/kb/on-device-models
On top of that, current mobile versions (and weirdly enough only mobile versions) can now also summarize articles using Apple Intelligence (where available) or Mozilla-hosted remote inference: https://blog.mozilla.org/en/firefox/shake-to-summarize/
I imagine google has to build something that works for basically every kind of user out there, vs what you built. Moreover, it’s self obvious that they would support Gemini but not ollama, again given most users cannot run beefy LLMs on their consumer devices.
https://news.ycombinator.com/item?id=45292260
https://news.ycombinator.com/item?id=45292163
They do have phone support, but they refused to unlock the account and just said she'll never be able to use primary email account with Apple's systems because of the frozen account.
So yes, any cloud provider can lock you out for arbitrary reasons. Just because they answer the phone doesn't mean the customer support agent can actually do anything about it.
Photos sync to iCloud is terrible slow though compared to Google Photos - syncing 100GB take days and 500GB takes forever. At least it end-to-end encrypted with Avanced Data Protection. But yeah if you multi-TB photo archive buying large storage options of iCloud make no sense simply because it's impossible to use.
I'd better use self-hosted Immich.
So, on the consent-quality-useful triangle (WIP), Google is clearly eliminating quality and consent to provide you with a useful interface to the Google consentless compression box. Just what everyone wanted. The future is now.
Notification: You have 2 new views (details button: 2 ad-consenting views, 0 other views) on the photo you took of the compression artifact on a video that you suspect Google might have accidentally compressed without your consent, confusing itself to be Instagram. Unfortunately, your comparison photo gets equally confused and is compressed to be equally as bad as the compressed one. Now the photos look identical, and you look like a conspiracy theorist tweeting about "video encoding" from your Sesame Street Elmo phone, just like everyone else, with no issue at all. "We're in the Ourobouros. Maybe Paramount isn't the issue. Maybe it's Paramount Plus." The Samsung rolly-polly bug interrupts and insists this issue will have to wait because it's 2pm on Friday. Now, your Elmo phone is now the only device still working in the office, as you try to convince your wife why you have to stay late, "Because you're different than the rest of the people posting compression artifact-laden photos."
https://www.theguardian.com/technology/2022/aug/22/google-cs...
Gosh I hate google products.
LLMs interacting with markup is not the best abstraction layer.
But you might want to be careful about which web pages you share this way?
That is why every AI company is making a browser.
> Have a question about what you're reading? Ask Gemini. It uses the context of your open tabs to provide relevant answers and explanations, keeping you focused.
In France some bits of the page are localized, some are still in English -- doesn't project professionalism or inspire confidence.
> To use Gemini in Chrome on your computer, you need to: Be 18 or over and in the US. Use a Mac or Windows computer. Use the latest version of Chrome. Learn how to update Chrome. Sign in to Chrome. This feature isn’t available in Incognito mode. Learn how to sign in to Chrome. Have Chrome’s language set to English (United States).
Why can't I set Chrome to whichever language I may want and still have that Gemini thing in english?
https://news.ycombinator.com/item?id=45292260
Maybe someone can post the change log tomorrow and we can do it again.
I'm thinking over the weekend we could post the GitHub merge of these AI features so we can give Google even more exposure.
By Tuesday I hope someone will write a review of these features rehashing the same thing. I'd love to have that be upvoted to the top of HN again.
It was forced into Windows task bar as well.
This seems to be in the same vein.
Typical corporate doublespeak. The web is neither "mine", nor am I ever in control. If anything, the web belongs to corporations like Google. By integrating their text prediction, summarization, and hallucination engine into their web browser, they're further cementing their position of control.
[0] https://drive.google.com/file/d/1Nx4gJA-qWodYWm-SK87Aa63i_jF...
My guess is it's either the first part (doing it via Gemini in Chrome) or the last part (permissions enabled).
chrome://settings/ai redirects to chrome://settings (general settings). Manually searching 'ai' brings up dozens of other settings - stuff like 'mail' (which contains 'ai' string) - but nothing Gemini-related.
On the most up to date chrome: Version 140.0.7339.186 (Official Build) (arm64)
The instructional video [0] says there should be a 'Gemini' icon on the top of the Chrome browser, but I don't have one (macOS). (do I have to have a paid Gemini account for it to be there?).
In any case, when OpenAI and Grok launch things, I usually just go and try them in about 20 seconds. By comparison Google's AI launches are tedious..
(thanks for the help btw)
I can confirm a paid Gemini account is not [corrected] needed.
No problem, it took me a minute to get it enabled myself - not sure why it's so special cased for what it is.
Google should say this up front (or at least prompt that I need to pay) rather than wasting users' time.
The 'How do I use Gemini in Chrome?' section of their launch page doesn't say anything about that requirement either.
Anyway, </rant>. Thanks for your help.
To reclarify: A paid account is NOT needed. I do not have one and it works. Do note the page says a US based Google account is needed though, it just doesn't need to be a paid gemini account.
https://drive.google.com/file/d/1ZifoxoUSy1vEgh2Qx8GaCsb5ywE...
Like if users can just get the info they want right at Google.com why would they click through to any of the search results? Isn't that stealing clicks from websites?
Google is about to break even further away in the LLM race with this move, seeing as they will be getting an absolutely, supremely stunning amount of regular and novel data 24/7. Not everyone uses dedicated LLM interfaces, but more people I know use Google search. As Google === Search for so many.
Nevertheless, it is an business savvy move to make, considering the recent ruling by the judge to not force Google to split apart or break up its business w/r/t to Chrome.
A monopoly is when you do anti-competitive things, not when your product is far and away the most popular.
If anything blame Firefox for dropping the ball so damn hard
Google is an anticompetitive monopoly, and Chrome is an anticompetitive monopoly. This has been established by multiple courts of law. Your armchair claims to the contrary hold no water.
The case ruled that google search was a monopoly for using anti-competitive practices (paying apple to make google the default search) and one of the recommended remedies was selling off chrome. The judge didn't go through with this, likely because in the case they note that people voluntarily really like using chrome, because "edge is the default windows browser that people seem to use only to download chrome".
Check the chair you are in before speaking.
Welcome to the Vibe Browsing security nightmare.
I'm surprised this didn't happen sooner... The amount of data available from Chrome users seems enormous.
I would like to inspect some part of the DOM and chat about it with an LLM, including the CSS rules that are applied to each subnode in my selection.
iOS locks you down to safari already.
Then web attestation and platform attestation/drm for mobile apps is eventually firmly in place and it means you can only use either android or ios for paying your bills (already a thing for most banks where I live...) or even doing mundane government bureaucracy.
god, what a timeline... and even if you don't live in the country responsible for this mess you still have to suffer these consequences and everyone is so apathetic and shuts their brains off when mentioning any of these problems.
Given this, i still won't use Chrome.
These are all things Apple could build into safari, but they're nowhere to be seen. They'll be stuck solving yesterday's problems (like building an infinitesimally better camera for the latest iPhone), but not at all integrating any AI into them.
Not to mention that actually giving Google money for anything other than an in-app purchase is oddly hard work - try buying a Google business subscription and behold an interface worse than AWS's console. Google has so much catching up to do that it's conceivable that they'll eventually fail.
Once Chrome integrates Gemini Live amd treats your browser as a video input stream, it's pixels all the way. No lag, no incorrect clicks on hidden elements.
Nobody asked to this. Interpreting websites for its users is categorically not what a web browser is for.
A local model capable enough to do the things that this is designed to do? Yes please.
Gemini in Chrome is a way to increase adoption. Gemma in Chrome is an innovation - a platform that allows developers to build stuff leveraging the local model. A step closer to a world where we can talk to our computers and have them do what we mean instead of what we say.
You didn't want it in your phone, bang, it's there!
You didn't want it in your browser, bang, it's there!
Next, coming to a fridge near you! /s
Well, with Samsung forcing ads on their "smart" fridges [0], Google + AI can't be far behind.
Thank god we have strong regulation in the US to protect us. /s
This period marks Google's transition from a preparatory phase to an aggressive market push, which is thus far yielding significant momentum.
This contrasts with the apparent friction in the Microsoft-OpenAI partnership, where long-term strategic alignment seems uncertain. Furthermore, there's a growing perception that competitors like Anthropic are achieving superior performance in specialized domains like software engineering. This suggests OpenAI's current model, which appears heavily focused on optimizing its existing architecture, may be approaching diminishing returns on genuine innovation.
