Fake Tinder Profiles: This Photo Is Not Real
It looks like a professionally taken photo of a beautiful young Japanese woman. Maybe you’ll come across her on Tinder and be immediately drawn to her. At the latest when she tries to avoid meeting up with you, you should become suspicious. Is she a catfish? Yes.
The truth is, that this photo is a fabrication of modern A.I. technology. This woman was never born. She never existed. Scary, isn’t it?
By now, A.I. is capable of creating these photos out of thin air. Here are a few more examples:
The technology behind these seemingly real images is a GAN (Generative Adversarial Network) called StyleGAN, developed by Tero Karras and his team of researchers working for Nvidia. A.I. uses deep learning to create images that the human eye can’t tell apart from real photos.
What does this mean for us? People could theoretically abuse these images to create fake stories about fake people, maybe in order to create heartwrenching stories for GoFundMe campaigns, to catfish people on dating sites, or to simply pretend they are someone they are not.
Since none of these people really exist, posing as them would not count as identity theft — there’s no existing identity to steal. If we fabricate fake news, these images could make them more believable.
You could have met someone online today that you believe is a real person, but turns out to be yet another fabricated identity. How would you know?
The short answer is: You can’t. Well, unless you decide to meet up and the other person refuses.
If you want to look at more images of fake people, you can do so on their website:
So the next time you meet someone on Tinder, you may want to ask yourself whether they are real or not.
The good news — Cats!
While the knowledge of such technology existing may have made you a little more paranoid when it comes to meeting strangers online, there is a (small) positive side effect too.
The tech can also create an unlimited amount of non-existent cats! You’ll never run out of cat pics ever again!
And if you want to fool your chemist friends, it can also create chemical compounds in 3D that don’t really exist either.
Much like DeepFake, this technology shows that A.I. can create things in seconds that would take us much, much longer, and fool us into believing that it’s real. Even worse, if those two technologies meet sometime in the future, it could create complex videos of people that don’t even exist.
How it works
For those of you who are into programming and machine learning, here are a few links describing how the technology works in detail:
- A 2-minute paper on Youtube explaining how StyleGan-v2 works
- A Youtube video explaining modifications made to the StyleGan architecture for faster learning and removal of artifacts
- The complete StyleGAN-v2 source code for you to create your own
It’s completely open-source. Which also means that it can easily find its way into the wrong hands. Just saying.
The true power of A.I.
A.I. is already taking over many parts of our lives, maybe without you even being aware of it.
Tesla’s autonomous driving is also based on deep learning. The long-awaited full self-driving has just been rolled out as a beta. And Nvidia has impressed gamers and 3D-artists alike with two of their latest technologies:
RTX, which allows real-time raytracing (a process that before was only possible with pre-rendering an image — often taking hours)
DLSS (Deep Learning Super Sampling), which can create sharp and high-resolution graphics while not losing any speed, making the gameplay fluent and high-fidelity without any compromises in graphics.
A.I. has already made incredible progress over the past 5 years. And with deep learning becoming more and more sophisticated, it’s a matter of time until we have the full A.I. you know from old science-fiction blockbusters.