OpenAI's New Social Video App Will Let You Deepfake Your Friends (theverge.com) 22
Alongside its updated Sora 2 AI video generator, OpenAI has launched an iPhone-only social app called Sora that lets users consent to have friends create deepfake-style cameos of them. The invite-only app works a lot like TikTok with short remixable videos but enforces restrictions on public figures and explicit content. The Verge reports: In a briefing with reporters on Monday, employees called it the potential "ChatGPT moment for video generation." The Sora app is currently only available to US and Canada users, with other countries set to follow, and when someone receives access, they also get four additional invites to share with friends. There's no word on when an Android version might be released.
Sora users can give their friends -- or, if they're feeling bold, everyone -- permission to create "cameos" with their own likeness using the new video model, which is dubbed Sora 2. The person whose likeness is being generated is a "co-owner" of that end result, OpenAI employees said, and they can delete it or revoke access to others at any time. Like TikTok, OpenAI's Sora app allows you to interact with other videos and trends using a "Remix" feature, but it only allows for the generation of 10-second videos for now.
Sora users can give their friends -- or, if they're feeling bold, everyone -- permission to create "cameos" with their own likeness using the new video model, which is dubbed Sora 2. The person whose likeness is being generated is a "co-owner" of that end result, OpenAI employees said, and they can delete it or revoke access to others at any time. Like TikTok, OpenAI's Sora app allows you to interact with other videos and trends using a "Remix" feature, but it only allows for the generation of 10-second videos for now.
A nurse wiping Ron Jeremy's ass every day. (Score:2)
There is more in life than just porn.
Sounds about right... (Score:3)
Ingenius! (Score:3)
OpenAI, the content producer rapists [slashdot.org], now allow you to virtually rape your friends' identities. Ingenius!
Re: (Score:3)
Re: (Score:3)
I forget, how many hundreds of $billions have gone in to building this nihilist trip-toy?
Enough to leave a crater the size of Saturn when this bubble implodes.
Re: (Score:3)
They made a piss-poor attempt to make it sound better with language like "Sora lets users consent..." What a weird thing to build, a consent app. How did we ever consent without it?
Protecting the grid for the win! (Score:3, Informative)
but it only allows for the generation of 10-second videos for now. [emphasis added]
Protecting (maybe) the electrical grid for the (sort of) win. [futurism.com]
The Futurism article says "the energy demands of text-to-video generators quadruple when the length of a generated video doubles" and "To spit out a five-second clip, the researchers found that it takes the equivalent of running a microwave for over an hour."
That's 4 hours of microwave oven use for 10 seconds.
Now, I'm not saying that Sora takes this much energy (given its use case, it's highly unlikely), but it very well could have the "double the video length, quadruple the energy use" problem, so a short time limit makes sense.
who needs friends of you can have OpenAI deepfakes (Score:3)
Only problem is who's pictures to use initially to generate those deepfakes.
DAAS what could go wrong? (Score:3)
Not creepy at all (Score:2)
<honestly how did we get here>
But some 10 year old boys probably have some great ideas for some short videos, I'd bet.
Because they'll be my friends forever (Score:3)
Re: (Score:2)
They'll never get pissed off,. break my phone, make a fake kidnapping video and extort money from my family. Just like they wouldn't murder me and automate sending convincing video postcards so everyone thinks I'm doing fine on my vacation.
I'm intrigued by your business model and would like to learn more.
Re: (Score:2)
My first thoughts were along these lines as well although a little less extreme. You might want t be a little more selective about your friends...
That said I basically agree. I don't see a case in 2025 world where I'd want to explicitly consent to even very old very trusted friends create high fidelity media content possibly difficult to distinguish from real recordings of me saying and doing things without my input.
As you say people have falling outs, sometime well meaning people just make bad decisions,
Re: (Score:2)
Most likely scenario, there aren't going to be any actual friends involved.
Step 1. Create an account. Upload some pictures of someone you have the hots for or videos or whatever you need for this, which you've probably downloaded off their social media.
Step 2. Create a second account. Mark the second account as a 'friend' of the first one.
Step 3. From the first account, give the second account permission to generate deepfakes.
Step 4. Generate away. This company claims they don't allow explicit content, whi
Two roads appear ahead (Score:2)
One leads to AI assistants that will help us solve previously intractable problems in science, engineering, medicine, etc
The other leads to AI generated "friends", AI generated ads, AI generated music and a tsunami of other AI generated crap
The value in AI friendships :o (Score:2)
“Or, ‘I need to have a hard conversation with my boss at work. How do I have that conversation?’ That's pretty helpful. As the personalization loop kicks
Re: (Score:3)
“Or, ‘I need to have a hard conversation with my boss at work. How do I have that conversation?’ That's pretty helpful. As the personalization loop kicks in and the AI starts to get to know you better and better, that will just be really compelling,”
And your boss can use an AI too, then the AIs can babble to each other and let you know later whether or not you've been fired.
What safeguards? (Score:2)
Digitize your friends for future advertisements! (Score:2)