15 comments

[ 3.0 ms ] story [ 36.8 ms ] thread
> “deepfakes” — videos made with artificial intelligence that fabricate a lifelike simulation of a sexual act featuring the face of a real woman.

What?? AI doesn't work on men?? ;)

Wieners are harder than hands. The robuts haven't figured them out yet.
Of course it does. That's why you have all those uncanny deepfakes with men's bodies and women's faces.

Theoretically I suppose you could do it the other way around also, if for some reason you wanted to.

If deepfake porn becomes trivial to generate, doesn't that by definition make all revenge and extortion porn worthless because now anybody can make porn of anybody else?

It seems like a natural continuation of being able to Photoshop people's faces onto strangers in compromising situations. Is this not just a better version of that?

Seems like fighting it would only make fake porn more valuable and dangerous.

> natural continuation of being able to Photoshop people's faces

This is where I'm at as well, where it differs IMO is: the throughput/variety and the zero barrier to entry—even the most gifted 'shopper could only work so fast and with whatever available media. It would be possible to deconstruct the image to know face from X, background from Y. And still, virtually all photoshops have "tells" — they're "off" with regard to lighting, perspective, etc. and it varies from artist to artist.

Tools exist that can tell which pixels were "disturbed" or otherwise inconsistent with the rest of the image. Being able to spot a Photoshopped image is an internet joke and true life skill—"it's a 'shop. I can tell by the pixels and from years of seeing 'shops"

AI-generated images removes the maker's mark, and despite hilariously struggling with anatomy, seldom gets lighting or coloring "wrong" in a way that a human might.

That's all sensible. My point still remains that when the cost, time and technical competence of generating fake porn of someone approach 0, and everybody knows it, then the value of whatever porn you might have of that person is also 0, which is the state we ultimately want to get to.

Same thing with voice, we're almost at the point where anybody can spoof anybody else's voice with barely any effort. We're going to get used to a world where voices mean nothing.

(comment deleted)
(comment deleted)
Or maybe it will become more scarce and "valuable" due to behavioral changes.

Eg, if having a bunch of photos of a person is tantamount to having a bunch of porn out there of them, maybe people will become much more protective of their image.

Maybe they will react to being photographed in public in the same way they'd react today if someone tried to stick a camera under their skirt.

Maybe public figures will all just accept that this is going to happen to them, but it doesn't follow that everyone else would accept it. Maybe photos, any kind of normal photo, will be seen as a sensitive personal item you'd never willingly release to the public.

It sounds strange to say people will no longer allow photos of themselves, but is that really more strange than saying everyone will accept hyper realistic AI porn of everyone is floating around?

Another bullshit article. Thanks to articles like this companies waste resources on "ai safety" which practically means the model refuses basic requests because "oh it might offend someone and some journalist might smear our company".

Meanwhile open-source models are already there and are be here to stay. So instead of trying to uphold the unsustainable status-quo that images are trustworthy, it's time to realize they aren't.

This is probably the reason I will stop paying for the stability API and just host my own instance of Stable Diffusion. I’m not trying to generate porn or sexy images, my problem is the massive amount of false positives! My project revolves around making video to video animations with Stable diffusion but when 10 images of your 100 frame sequence is blurred out (for safety) it ruins the entire thing, you’ve lost both time and money. I complained to Stability and they were kind enough to refund me some credits.

The funny thing is an image of someone winking at the camera will trigger it but a bunch of unsolicited nipples makes it through with out any issues.

It’s too bad because I want to support Stability but these false positives will force me to stop paying them directly for the API.

I f*** hate it when tech companies try to play the role of moral authorities. Just because some journalist tries to score easy click points talking about porn.