Hacker News Reader

Comment by anju-kushwaha | Hacker News Reader

51 pointsanju-kushwaha18 comments

This is a blog post, not a Show HN. https://news.ycombinator.com/showhn.html

No, this is a warning of the current sloperation and data exfiltration by the "Dont. Be evil" folks.

Sorry but it’s not really an opinion for you to contest it, it’s a well defined guideline of what makes a Show HN post vs a normal link share. If the OP wants to add commentary to their submission they could use “Tell HN” or simply point at the actual blog post and write a comment.

This, and "don't be evil" was replaced over ten years ago: https://en.wikipedia.org/wiki/Don%27t_be_evil

So stop calling the Google people out for being evil. They are officially evil since 2015.

barring fully local AI, not sure how it can work without using your data, by design. it's personalized or not.

People are not concerned because it can, because as you say it needs to in order to operate if you opt in.

The concerns implied are that:

* You are automatically opted in, or you are nagged to opt-in in a manner that makes it easy to accidentally do so, as is not uncommon with these things.

* That the use of your data for training isn't just for you, it is also training any other models Google wishes to use it for.

* That once they have started, opting out does nothing: your data is already theirs. Not that wasn't already anyway, control of how your stuff is used is apparently a right reserved only for corporations.

You could instead use a trustworthy LLM API provider instead of an American one bound by legislation like the US CLOUD act.

Confidential compute is another option.

The EU angle is interesting but the more immediate concern is the opt-in/opt-out framing. Google historically defaults these features to on and buries the toggle — did they actually make face data access opt-in from the start, or is this another "we'll notify you and assume consent" situation? The GDPR enforcement gap between announcement and actual compliance has been wide enough to drive a truck through.

According to the article, you can fix this by turning off Gemini Apps Activity in https://myactivity.google.com/product/gemini

I already have this turned off. It's slow, but I use a browser extension to save my Gemini chats locally, when I want to keep them.

Do they train Gemini other people use based on this or just generate images for me?

Unless something changed I'm not aware of, you have to opt-in to Personal Intelligence for private data. It asked me about 10 times to turn it on, I said no each time, it's still not enabled.

If it asked you 10 times, that's 10 chances for you to mis-click.

"Opt-in" is a lie when the user is harassed or tricked into "accepting".

idk why but Gemini can't even read gmail attachments

I uploaded an image to google maps once on my iPhone. A few weeks later I get a notification - "How was City Park? Care to share a photo with Google Maps?". I noped out pretty quick. I'm 99% sure it asked for permission to upload a photo and did not explicitly describe how it would be scanning all of my photos in the background.

Creepy and gross.

Moved all my photos over to Ente recently, and I'm very happy to no longer be feeding all of that data into the Google machine. I'd highly recommend everyone else move their photos out as well (Immich is another good option).

I wonder if it will be claimed one day that the AI got snoopy and didn't listen.

"Snooping as usual, I see!"

Someone just handed me a smartphone today; the second thing I saw it say was something to the effect of "Gemini can look at your screen now (CoPilot outrage style)" and then I politely noped the fuck out of the building.

I expect that info to be completely sufficient to dox me; in the meantime, while waiting for the boys to arrive -- does anyone know how to make the camera in Waydroid not crash the entire stack? Or am I to understand that as yet another bloody rite of passage?

Reply on news.ycombinator.com