this post was submitted on 19 Jul 2024
226 points (97.1% liked)
Privacy
31837 readers
90 users here now
A place to discuss privacy and freedom in the digital world.
Privacy has become a very important issue in modern society, with companies and governments constantly abusing their power, more and more people are waking up to the importance of digital privacy.
In this community everyone is welcome to post links and discuss topics related to privacy.
Some Rules
- Posting a link to a website containing tracking isn't great, if contents of the website are behind a paywall maybe copy them into the post
- Don't promote proprietary software
- Try to keep things on topic
- If you have a question, please try searching for previous discussions, maybe it has already been answered
- Reposts are fine, but should have at least a couple of weeks in between so that the post can reach a new audience
- Be nice :)
Related communities
Chat rooms
-
[Matrix/Element]Dead
much thanks to @gary_host_laptop for the logo design :)
founded 5 years ago
MODERATORS
400-700 for a single article of clothing with no mention of what facial recognition software this affects, how effective it is and what is the failure rate, error bounds, etc. Sounds like a scam.
I wouldn't call it a "scam" just manipulative marketing. This stuff doesn't seem like it'd work for any of the modern facial recognition options, but that's just a guess. If it did work well and they were proud of it, you can be sure that'd be part of the marketing, so it at best is mediocre if not useless.
So I don't know if you guys actually read the article or not but they absolutely DO claim that it works against YOLO which they claim to be the most popular recognition software. I don't know about how factual any of that is, but they do make the statement.
I wouldn't call it a "scam" just manipulative marketing
the difference?
Not who you asked, but I think some might argue that it would be a scam if you ordered it and it didn't arrive or something like that. If it works against one facial recognition model than technically it is just bad marketing. Either way is bad, though.
Unfortunatly its a cat and mouse game. Except the cat is a easily deployable software problem and the mouse is buy new clothing hardware problem.
Led clothing anyone?
they've been around for some time now: https://www.reflectacles.com
Ghost uses a frame-applied material that reflects both infrared and visible light. In low light environments they will maintain your privacy on cameras using infrared for illumination and also block 3D infrared facial mapping during both day & night. The visible light reflection can make you anonymous in images/videos using a flash in low light.
Ghost is $170 (US, I'm assuming). Not great but not bad for a wicked cool looking pair of sunglasses. Considering Ray-Bans are around $200 (and, no offense, look like they're from Tesco), and that Ghost are privacy focused, I'd say that price seems not that bad. Still high, though.
load more comments
(4 replies)
If they can target the underlying architecture of the models like nightshade does, it will actually be quite hard to deal with for the surveillance companies.
Interesting concept if we can target and poison the definatly stolen training data.
The method that Cap_able has patented allows the wearer to incorporate the algorithm into the fabric of the clothing and still look stylish.
I was with you up until the stylish bit
Everyone’s a critic
I for one think looking like a texture ripped from DooM is stylish.
load more comments
(6 replies)
This will work for about 10 minutes. Better off wearing a facemask, bandana, juggalo face paint, etc
can't they still id you with a mask? apples face id can work with a mask. though I suppose that has depth data too
Almost certainly. The facial ID is good enough that US customs didn’t even want to see my passport. Just a photo was enough to let me back in the country. I even significantly changed my hair between departure and arrival. Shit is scary.
It's only a matter of time before a cop charges someone with obstruction for trying to disrupt a camera system (during the commission of a crime, I mean).
Or they just work around it
So I guess we're wearing broken JPEGs now huh?
I want this to be a thing
Glitch art clothing would be dope even if it didn't help with AI fuzzing.
Good for privacy! But I really doubt it would work for all recognition systems.
Some funny pitfalls that may occur - Self driving cars would prefer to hit that person if had to make a choice between him and some other human. And, there is possibility that the Street mapping cars would not blur his face for the lack of detection.
AI probably was already patched 5 minutes after the article came out.
You can't really "patch" LLMs like most software; you'd have to retrain them, no?
Yeah but they don't use LLMs for this, they'll use some other kind of machine learning mixed in a big pipeline of data processing. It makes it really hard to guess how much work it would take to fix. It might require retraining, might just require an easy patch of the rest of the pipeline.
My guess is that they're just shitty jumpers and there's nothing to fix anyway.
load more comments
(2 replies)
$246?! I can't afford that. For that price I'd rather avoid cameras and such. Cool technology though
Those people are just dressed like regular Australians.
🎶"Because I'm tacky..." 🎵
This would be a good article if the pictures actually showed people wearing the clothes.
Literally the header image...
What's with the floating heads?
I see a couple people, and some oddly colored blobs.
Oh! HHahhahhhHah! That's a good joke! Wooshed right over my head hahahahahahahah!
Edit: correct autocorrect
William Gibson's Ugly Shirt come to life
Absolutely cool. I will have to revise all my internalized cyberpunk imagery though.
Their demo video looks horrible. They are using a trash algorithm to demo the detection failing.
The girl also moves extremely slowly and permanently has her arms out to the side at the elbows. I assume this is the only way they could get the results they wanted to show.
view more: next ›