Kids "easily traceable" from photos used to train AI models, advocates warn.
I mean, that's true, and could be a perfectly-legitimate privacy issue, but that seems like an issue independent of training AI models. Like, doing facial recognition and such isn't really new.
Stable Diffusion or similar generative image AI stuff is pretty much the last concern I'd have over a photo of me. I'd be concerned about things like:
Automated inference of me being associated with other people based on facial or other recognition of us together in photos.
Automated tracking using recognition in video. I could totally see someone like Facebook or Google, with a huge image library, offering a service to store owners or something to automatically identify potential shoplifters if they let them run automated recognition on their store stuff. You could do mass surveillance of a whole society once you start connecting cameras and doing recognition.
I'm not really super-enthusiastic about use of fingerprint data for biometrics, since I've got no idea how far that is traveling. Not the end of the world, probably, but if you've been using, say, Google or Apple automated fingerprint unlocking, I don't know whether they have enough data to forge a thumbprint and authenticate as you wherever else. It's a non-revocable credential.
Like, I feel that there are very real privacy issues associated with having a massive image database, and that those may have been ignored. It just...seems a little odd that people would ignore all that, and then only have someone write about it when it comes to running an LLM on it, which is pretty limited in actual issues that I'd have.
And all that aside, let's say that someone is worried about someone generating images of 'em with an LLM.
Even if you culled photos of kids from Stable Diffusion's base set, the "someone could generate porn" concern in the article isn't addressed. Someone can build their own model or -- with less training time -- a LoRA for a specific person.
kagis
Here's an entire collection of models and LoRAs trained on a particular actress on Civitai. The Stable Diffusion base model doesn't have them, which is exactly why people went out and built their own. And "actress" alone isn't gonna be every model trained on a particular person, just probably a popular one.
And that is even before you get to various techniques that start with a base image of a person, do no training on that image at all, and then try to generate surrounding parts of the image using a model.
Even if you're not on social media, you'll probably still have a shadow profile on Google/Metas servers.
My 13 month old baby has a library of images searchable in Google photos and a profile photo in the app. It's convenient, but incredibly creepy.
He said that it's creepy but convenient, digital privacy and laziness don't go hand in hand generally. Every week I read about another alternative for Google Photos, so the solution is not far away (three posts down I found this for example). To each their own I guess, but with such simple solutions I can't justify using Google's spyware.
It's not opt-in as far as I'm aware. Just using Google photos makes it so.
I suppose I'm deep enough in the google ecosystem (well, let's say my wife is not going to move away from it) to be desensitised to how messed up it kind of is.
I was more talking about how other people (i.e. your friends) will take photos of you and post it on social media or even just keep them in their google photos, and meta/google will build a shadow profile for you without your consent via facial recognition.
No, but it's opt-out, and it is your responsibility to ensure that stuff like this doesn't happen - full disclaimer, that is my personal opinion. Pictures of third parties that did not give explicit consent for each and every picture shouldn't be uploaded to cloud providers etc., let alone pictures of kids and other parties who are unable to give proper consent.
My wife is incredibly careless with these things. She wants to know how to properly operate her smartphone and wants to care about e. g. privacy, and on paper, she does - but in practice, we do a 2 hour long session, I explain all the settings to her, where to find them, why they are important, what implications certain actions / options have for security, safety and even keeping her phone in working order, yet as soon as she walks out the door, she no longer cares one bit, will blindly click to accept all kinds of EULAs and default options, never investigate what the notifications about failed backups mean, never delete obsolete / already backed up data etc. up to a point where her phone no longer works and she then instructs Google Photos to upload multiple years of family pictures full of private moments, multiple children etc. to Google.
The UI is crappy enough so you'll spend a significant amount of time deleting the pictures remotely, absolutely infuriating. I was furious, in particular because I can't say that removing the pictures will also reverse all the potential consequences of sharing all your pictures with Google.
For reference, Google Photos does offer facial recognition, stores and estimates locations and even estimates activities based on media content.
IMHO, being this negligent is not excusable in this day and age.
I agree with you mostly, and thank you for giving such a passionate and important response.
The problem is not the people though. Placing the "blame" or responsibility on the victims of this invasive behaviour is not the correct conclusion. These settings are deliberately obfuscated and people are uneducated on privacy and how it relates to technology. This is not their fault. Life is far too complicated to place yet another burden on the individual who already has so much to think about. The change needs to come from the people, yes, but it is not the people who need to change.
You are correct. It was probably not perfectly clear from my response, but I do not want to blame the individual here.
Naturally, the "Backup all my files" setting should not be opt-out, and when opting in, there should be easy and succinct explanations of what the implications are.
Lemmy as a whole is apparently a very technical community, so we often tend to forget that an understanding of these implications does not come naturally to all users, and that there are people that need a phone just like everyone else, but might not be in a position to acquire the knowledge required to make an informed decision.
I am fully with you regarding your conclusion, up to a point where I applaud regulatory action that protects customer interests, including privacy. I do not believe that companies will sort out these problems (or in any form of liberal "self regulation", really) on their own, since it's not in their interest to do so.
I guess I wanted to express that while things are obfuscated and software is full of malicious anti-patterns, we do have to take extra care to protect ourselves, and, as was the topic here, our kids. I still actively try to work on changing the current status though, politically or by making political decisions, e. g. looking at open source / projects that are more aligned with what I'd consider to be in the best interest of users, and I'd encourage everyone to do the same.
Wait until you have photos spanning from, not only your child, but your cousins' children who are photographed less often. Google can easily match up an infant to the same 10 year old child. Hell, I can barely do that sometimes and have to use context clues to figure out who the infant was.
I scanned a ton of my mom's family photos after she passed, and uploaded them to Google Photos. It's a bit shocking how good it is at guessing the same person at different ages, even 20+ years' difference.
Photos of Brazilian kids—sometimes spanning their entire childhood—have been used without their consent to power AI tools, including popular image generators like Stable Diffusion, Human Rights Watch (HRW) warned on Monday.
The dataset does not contain the actual photos but includes image-text pairs derived from 5.85 billion images and captions posted online since 2008.
HRW's report warned that the removed links are "likely to be a significant undercount of the total amount of children’s personal data that exists in LAION-5B."
Han told Wired that she fears that the dataset may still be referencing personal photos of kids "from all over the world."
There is less risk that the Brazilian kids' photos are currently powering AI tools since "all publicly available versions of LAION-5B were taken down" in December, Tyler told Ars.
That decision came out of an "abundance of caution" after a Stanford University report "found links in the dataset pointing to illegal content on the public web," Tyler said, including 3,226 suspected instances of child sexual abuse material.
The original article contains 677 words, the summary contains 169 words. Saved 75%. I'm a bot and I'm open source!
