Google's Privacy Policy Now Admits to Collecting All Your Data for AI Training | Hacker News Reader

Google's Privacy Policy Now Admits to Collecting All Your Data for AI Training

pcgamer.com

144 pointssizzle24 comments

Discussed earlier this week: https://news.ycombinator.com/item?id=36586170

I wonder if Gmail data is considered publicly available, given that Google has argued in court that “a person has no legitimate expectation of privacy in information he voluntarily turns over to third parties.”

If so, how long before we can put something like “Jack Smith lives at:” or “Jane Smith has a health condition:” into Bard and it will democratize doxxing for us?

It’s very important to know where the creepy line Google doesn’t cross is these days. Is publicly available everything that the user communicates to someone else (shared files, photos, sent emails)? Is it what’s on the hypertext internet (in that case, the EULA is useless because you don’t sign Google’s EULA when you host websites unless you use Google Analytics)? Is all advertising data Google has on us considered public because it is already indirectly revealed to third parties through ad tools, or even all data Google has about us, period, because they said the expectation of privacy for that is unreasonable in court already.

There needs to be more clarity in these articles.

I doubt that Gmail data is considered as publicly available. Access to it is probably restricted even to most Google employees. Or else we would have heard about such gaps, just like we heard of Amazon and Tesla employees access customer's audio/video data.

Ads data is fundamentally different, as it's derivative from user data.

The absence of evidence is not the evidence of absence.

Truisms aside, in many cases the absence of a thing which one would expect to see is in fact a support for alternate hypotheses.

ie. "Given a world X is true, I would expect to see Y. Since I don't see Y, that's at least weak evidence that X may not be true."

There are of course plenty of other reasons why you wouldn't see Y, but it's at least some kind of signal.

But it doesn't necessarily have to be considered 'public data' for Google to be able to exploit it. Google uses opt-in by default for privacy settings and has a long track record of being misleading when it comes to privacy. I'm sure there's plenty of people out there who have unknowingly opted in to 'volunteering' their Gmail data or will soon be.

On the other two threads, many people considered Gmail to be "public" because it wasn't end-to-end encrypted.

Google sucks, but that is mostly because their products suck.

The war for privacy is lost. If you want to be private online, you don't just log onto your computer, connect to a VPN, and install a new web browser... You need to buy an entire new computer, go to a public wifi spot with a full facemask and... oh you gave away your location.

Online privacy is a farce that pretending exists only puts people at risk.

Companies have billion dollar reasons for finding out who you are, and its not just for ads. Some want to prevent bad actors. Some want to prevent scraping. There are entire companies whos job it is to identify you. Are you, 1 person, able to beat a multi-million dollar company who's job it is to identify you? Oh and once you are identified once, you are added to some list which makes it easy to ID you in the future.

Well looks like the house is on fire. No sense trying to put it out now. Besides fire prevention doesn't stop fires so telling people fires are preventable is just putting them at risk.

As if ads wasn’t AI training. Ranking recommendations and coefficients rely on the same mechanisms.

It’s really hard to make the case here that there is anything new happening here.

Just the chickens of decades of not engaging with the topic coming home to roost.

> It’s really hard to make the case here that there is anything new happening here.

LLMs have more potential for weaponising personal information against us. They will comply with anything requested in the prompt.

you think ranking classification couldn’t weaponize ? like credit scores? search engine results ?

They didn't say that. They just said this has more potential than that, which does seem right.

> LLMs have more potential for weaponising personal information against us.

Doesn't follow from

> They will comply with anything requested in the prompt.

LLMs are impressive in how they are able to produce plausible sentences, but they don't actually bring any new data correlating or collecting capabilities to the table.

"Naturally, Google collects data from your online activity, like the stuff you search for, the videos you watch, the things you buy, and the people you talk to, and the location data accessed through your Android mobile device. But "in some circumstances," it also collects information from "publicly accessible sources": If your name appears in a local newspaper article, for instance, Google may index the article and then share it with people searching for your name."

The first swathe is why I try to practice as much online privacy hygiene as I can. The second I why I'm fine with a relatively common name.

Has anyone been able to determine the exact bounds of what "publicly available" means? The previous two comment discussions on this topic have devolved into arguments on ethics, and not really answered that question. All I was able to determine was, some people consider "public" to be outward-facing (I.e., something like Gmail would not be used for training), whereas others claim that anything that is not e2ee will be considered "public". That's no real help.

Publicly available data. Clickbait.

publicly available in Google's privacy policy is the actual bait.

It's a rather unique perversion of a "Privacy Policy", as it has been pointed out: normal companies have a Privacy Policy for data you give to them in the course of using their app or site.

Google is now applying their "Privacy Policy" to everything they can possibly get to, anywhere on the Web; in other words, it applies to all the data they intend to take from us, whether we choose to give it up or not.

This is wholly unsurprising, given their track record of privacy before, as well as their cavalier attitude to copyright, scanning, and republication of every book in existence.

Do people expect that their publically posted data (like hackernews posts for example) are not scooped up by a multitude of bots whether or not it’s mentioned in the bots’ privacy policies?

I think there’s a big expectations gap between what people think should happen and what does actually happen.

Just because I post on publicly accessible websites does not mean I’m okay with random companies slurping up the information to use for their own purposes.

Yeah, truthfully, you guys are right. In the end, anything on the public Internet will get slurped up by a lot of shady, janky players. I've seen whole forums get mirrored onto some place in China for God-knows-what reason.

Google is, thankfully, operating in these USA and regulated by US law. And they're being somewhat up-front about their intentions. That's more I can say for 95% of the web crawlers active today.

what will happen when they only have each other to mine for data?

There are first studies on this topic: "THE CURSE OF RECURSION: TRAINING ON GENERATED DATA MAKES MODELS FORGET"

https://arxiv.org/pdf/2305.17493.pdf

Reply on news.ycombinator.com