1015
Absolutely not training data no way no sir
(i.redd.it)
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 29 Mar 2024
1015 points (98.0% liked)
Curated Tumblr
4096 readers
5 users here now
For preserving the least toxic and most culturally relevant Tumblr heritage posts.
Image descriptions and plain text captions of written content are expected of all screenshots. Here are some image text extractors (I looked these up quick and will gladly take FOSS recommendations):
-web
-iOS
Don't be mean. I promise to do my best to judge that fairly.
founded 2 years ago
MODERATORS
Google photos and apple have been doing it for years too, they’re like we found this person 50 times in your photo collection, why don’t you name them?
Apple, afaik, used to be doing this on-device rather than in the cloud. Not quite sure about the situation today.
thats if u trust them
I don’t. Corps gonna corp, if they can. But I’ve checked this using all the development, networking, and energy monitoring tools at my disposal and apple’s e2e and on-device guarantee does appear to hold. For now.
Still, those who can should audit periodically, even if they’re only doing it for the settlement.
Hero
Thanks
Security is in my interest, but yw
Heard. Today you tomorrow me.
Still on device for Samsung, not sure about others https://www.samsung.com/us/account/privacy-policy/#:~:text=Information%20Stored%20on%20Your%20Device%20Not%20Accessible%20to%20Samsung
They were inferencing a cnn on a mobile device? I have no clue but that would be costly battery wise at least.
They’ve been doing ML locally on devices for like a decade. Since way before all the AI hype. They’ve had dedicated ML inference cores in their chips for a long time too which helps the battery life situation.
It couldn’t quite be a decade, a decade ago we only just had the vgg — but sure, broad strokes, they’ve been doing local stuff, cool.