I think the TechCrunch headline is slightly more accurate than the Verge headline, which is "Facebook is starting to feed its AI with private, unpublished photos".
In both cases they imply training models is happening when that's not been confirmed.
(Facebook could help here by answering press inquiries about it, which they apparently have not done.)
Yeah, I read the title of the other one and thought it was just about pictures you sent through Messenger or put in private on Facebook, so I wasn't too bothered (because I assumed they'd do this already), but actively reading your camera roll is next level.
but actively reading your camera roll is next level.
Now that it was established that they wrote malware to bypass tracking protections, nothing surprises me. Apps written by Meta are malware, as far as I'm concerned.
While I agree the articles are click bait since this as not been confirmed, it's not far fetched to assume that a giant corporation with a terrible track record and a big legal department worded their TOS like this because they intend to use that capability.
Given Facebook is about as voracious an actor of surveillance as has ever existed, their track record of respecting few red lines until they have been caught crossing them egregiously, and the bright spotlight Zuckerberg is currently shining on their AI ambitions, it defies reality to imagine them forgoing any data they can get there hands on.
You just know there is a dashboard that summarizes all potential data sources, and engineers wake up with the shakes and sweats, after dreaming that Zuck was standing behind them, with furrowed brow, and pointing to a stat that shows 2% of someone’s most private information still hasn’t been plundered.
Ok, a little hyperbolic. But he & Meta are relentless.
Comments
There's a big active thread about this here already: https://news.ycombinator.com/item?id=44401406
I think the TechCrunch headline is slightly more accurate than the Verge headline, which is "Facebook is starting to feed its AI with private, unpublished photos".
In both cases they imply training models is happening when that's not been confirmed.
(Facebook could help here by answering press inquiries about it, which they apparently have not done.)
Yeah, I read the title of the other one and thought it was just about pictures you sent through Messenger or put in private on Facebook, so I wasn't too bothered (because I assumed they'd do this already), but actively reading your camera roll is next level.
Now that it was established that they wrote malware to bypass tracking protections, nothing surprises me. Apps written by Meta are malware, as far as I'm concerned.
While I agree the articles are click bait since this as not been confirmed, it's not far fetched to assume that a giant corporation with a terrible track record and a big legal department worded their TOS like this because they intend to use that capability.
Exactly.
Given Facebook is about as voracious an actor of surveillance as has ever existed, their track record of respecting few red lines until they have been caught crossing them egregiously, and the bright spotlight Zuckerberg is currently shining on their AI ambitions, it defies reality to imagine them forgoing any data they can get there hands on.
You just know there is a dashboard that summarizes all potential data sources, and engineers wake up with the shakes and sweats, after dreaming that Zuck was standing behind them, with furrowed brow, and pointing to a stat that shows 2% of someone’s most private information still hasn’t been plundered.
Ok, a little hyperbolic. But he & Meta are relentless.
It is safest to assume that your photos are being used for training.