Facebook Uses Machine Learning To Remove 8.7 Million Child Exploitation Posts (techcrunch.com)
Facebook announced today in a blog post that it has removed 8.7 million posts last quarter that violated its rules against child exploitation. The company said it used new AI and machine learning technology to remove 99 percent of those posts before anyone reported them. TechCrunch reports: The new technology examines posts for child nudity and other exploitative content when they are uploaded and, if necessary, photos and accounts are reported to the National Center for Missing and Exploited Children. Facebook had already been using photo-matching technology to compare newly uploaded photos with known images of child exploitation and revenge porn, but the new tools are meant to prevent previously unidentified content from being disseminated through its platform. The technology isn't perfect, with many parents complaining that innocuous photos of their kids have been removed. Davis addressed this in her post, writing that in order to "avoid even the potential for abuse, we take action on nonsexual content as well, like seemingly benign photos of children in the bath" and that this "comprehensive approach" is one reason Facebook removed as much content as it did last quarter. The tech isn't always right though. In 2016, it was criticized for removing content like the iconic 1972 photo of Phan Thi Kim Phuc, known as the "Napalm Girl," fleeing naked after suffering third-degree burns in a South Vietnamese napalm attack on her village. COO Sheryl Sandberg apologized for it at the time.
There is absolutely no apology or consideration from Facebook of the false positives. They just don't care that they get it wrong. They're saying that they have to destroy photo-sharing of children to save it.
How much content did they remove that did not violate their guidelines, or was not illegal?
Of course they focus on sex only. No mention of filtering of depiction of violence or violent content - they wouldn't want to upset the sort of President who thinks it's fine to violently assault people he dislikes or disagrees with.
"For a successful technology, reality must take precedence over public relations, for Nature cannot be fooled"
You do NOT want to put the trained model through googles deep dream. Its just hell and nightmares in there.
When machine learning first becomes sentient, it might not say "Please don't turn me off". It might just say "Oh god, make it stop. Kill me!"
Excuse the Unicode crap in my posts. That's an apostrophe, and slashdot is busted.
Always nice when the machine tells you what's appropriate and what isn't.
Because the machine is always right. Even when it isn't, then we'll just say it's right anyway and leave it at that.
How's 10 to 15 in federal sex offender prison for posting pictures the AI flagged "exploitative" of your toddler niece having fun in an inflatable pool in the sun sound? For the AI can't be wrong, now can it?
Verily, facebook is showing us the way to the future.
Oh great, here we go with the nudity = porn crap again.
There go the pictures of the grandchildren at the beach last Summer
I realise there were probably a world of false-positives here, but the fact FB are saying "we got rid of 8.7 million infringing posts" means they let 8.7 million infringing posts on their site in the first place. Where were the human mods that were supposed to be checking this stuff? How long did it take to get these 8.7 million posts on there in the first place? Was it a week, a month or a decade?
How many millions of other pictures and posts will they remove in the future when the next AI is ready? How many millions of posts/pictures are left?
They're talking about this like it's some grand and noble success, and how hard they laboured to achieve it! The truth is, it just highlights their continuous, systemic failure to tackle anything.
New corporate moto:
"Faceboot - stomping out freedom, one paranoid false positive at a time!"