Facebook is Making Its Own Deepfakes and Offering Prizes For Detecting Them (techcrunch.com) 21
An anonymous reader shares a report: Image and video manipulation powered by deep learning, or so-called "deepfakes," represent a strange and horrifying facet of a promising new field. If we're going to crack down on these creepy creations, we'll need to fight fire with fire; Facebook, Microsoft, and many others are banding together to help make machine learning capable of detecting deepfakes -- and they want you to help. Though the phenomenon is still new, we are nevertheless in an arms race where the methods of detection vie with the methods of creation. Ever more convincing fakes appear regularly, and though while they are frequently benign, the possibility of having your face flawlessly grafted into a compromising position is very much there -- and many a celebrity has already had it done to them.
Facebook, as part of a coalition with Microsoft, the Partnership for AI, and several universities including Oxford, Berkeley, and MIT, is working to empower the side of good with better detection techniques. "The most interesting advances in AI have happened when there's a clear benchmark on a dataset to write papers against," said Facebook CTO Mike Schroepfer in a media call yesterday. The dataset for object recognition might be millions of images of ordinary objects, while the dataset for voice transcription would be hours of different kinds of speech. But there's no such set for deepfakes.
Facebook, as part of a coalition with Microsoft, the Partnership for AI, and several universities including Oxford, Berkeley, and MIT, is working to empower the side of good with better detection techniques. "The most interesting advances in AI have happened when there's a clear benchmark on a dataset to write papers against," said Facebook CTO Mike Schroepfer in a media call yesterday. The dataset for object recognition might be millions of images of ordinary objects, while the dataset for voice transcription would be hours of different kinds of speech. But there's no such set for deepfakes.
Warning to all camwhores: (Score:2)
Artificial Intelligence is coming for your jobs!
Re: (Score:2)
Re: (Score:3)
And that's assuming FB isn't actually using the human-powered detector to directly provide training feedback to their own deepfake generating system. Bcause they'd never be tempted by the myriad profitable ways that could be leveraged.
So, given that "perfect" deepfakes are all but inevitable, and probably in the near future - how can we as a society possibly deal with that? Investigative journalism is almost dead, and if video evidence becomes easily faked, what are we left with?
other side of the coin... (Score:4, Insightful)
How can they improve the creation of deep fakes so they can claim something real is a deep fake.
And they are asking for your help to do so,
Like water, Technology doesn't care how it is used.
and then there is not a coin, the bottomless pit of deception we all need to grow so to make it obvious enough that people will start to think for themselves instead of being lead by what just sounds good.
Re: (Score:3)
I remember an interview of John Lennon once, where he mentioned that (and I'm not quoting him here) the greedy people of the Earth don't want to control people, businesses or the economy - they want to control the whole damn universe.
I see these things, like that. It seems that the greed of the times is reaching for control over reality itself, or at least the ability to claim what's real and what's not, regardless of anything that everyday people experience.
That seems to be the place of greed. But the pr
Re: (Score:2)
How can they improve the creation of deep fakes so they can claim something real is a deep fake.
The funny thing is that GANs that are the leading way to generate fakes use a generator and a discriminator, where the discriminator's job is to spot the fakes and train the generator saying this doesn't look like real objects. So build a better discriminator and you can make better fakes...
Ummm.... is anybody really dumb enough to trust? (Score:4, Interesting)
Training AI to detect deepfakes goes hand in hand with training AI to generate them with high reliability and FB is in the perfect position to influence entire nations and mass populations with clever and subtle use of faking technology. In a democracy that is also all you need to ensure no threat of consequences.
Unfortunately they also have access to an insane amount of human eyes and efforts that can all this training for them, something the small groups working on this type of technology up to now haven't had.
Let this happen and that is it, end game.
Re: (Score:2)
Just in time for the 2020 election, to boot.
Hmm... Maybe Trump's bizarre behavior is a way to innocculate his base against deep fakes... he's
always doing stupid things, so nobody even pays attention to the deep fake of him masturbating
in front of a heaping platter of cheeseburgers.
Though... if someone were to make deep fakes of Trump apologizing for all the shit he's done and
admitting he was wrong about his weird conjectures... and generally acting like a president.... hmmm....
I saw a picture of Zuckerberg's soul on Facebook (Score:2)
So they ran out of ideas (Score:2)
What's next, detect which bits are pirated?
The shark (Score:2)
Where this is going (Score:2)
About 20 years ago, a friend in my writers group wrote a series of stories about a movement called "Directionists". They wouldn't believe anything they hadn't seen themselves, in person, with their own eyes. It's been a while, but I think the group came about due to something very much like deepfakes, though he didn't have that word.
He had a number of stories published, in Analog and Clarke's World, but I don't know if any of them included the Directionists.
Smart move (Score:2)
All of facebook is fake (Score:2)
what do I win?
Yeaaaaaah riggggggggght. (Score:1)
Crypto camera (Score:3)
will banks steal from your account as test (Score:1)
Re: (Score:1)
Old news (Score:2)
Zuckerbot attempting to appear human is the biggest deep fake of all.
In other news... (Score:2)
Platform uses forging tech to slide Overton Window after public outcry protesting the use of tech to create forgeries slides Overton Window on that platform.