Gone the days of photo shopping where fake images only remained as a joke. The advancement in this technology with empowering tools such as AI has everyone worried. Recently, a lot of pornographic images and videos of Hollywood actresses and famous celebrities which are highly convincing, have been circling over the internet and millions of people have been sharing the same. These obnoxious images and videos were later found to be forged and were declared fake by forensic experts.
Social media, the bearer and broadcaster of fake news too is posing a threat since the news it upholds spreads like wildfire and leaves little to do when the news is a hoax or a rumor. Remember what happened during #pizzagate? Day by day its what we see vs what we believe.
The video where Barrack Obama was seen talking rather unusually and calling Donald Trump a “Dipshit” also went viral, only to learn that it was American actor Jordan Peele just playing around with Artificial Intelligence technology. What is this technology that is so make-believe that we often forget to comprehend if it’s actually authentic? Today we’re going to explore the AI that has taken the internet by the storm and is seen as a potential threat to the near future. Deepfakes.
What Are Deepfakes?
Deepfakes are images or videos created using artificial intelligence where you can create real-looking forged images such as morphing popular faces onto pornographic sites. Some amateurish hobbyists started this on Reddit and this gave birth to the pioneering of so-called deepfake learning.
Deepfakes aren't just fake images or videos. Some are made with so much precision that it is really daunting to distinguish between a real and a fake. Deepfakes can be created using voice modulations too. So if you have seen a video of Nick Cage appearing in ten different movies he wasn’t a part of, or a politician saying rather unruly things, you have come across a deepfake.
Origin of Deepfakes- GAN
A general adversarial network-GAN is a deep learning technique in AI. Usually, machine learning is a hand-down job where you instruct the computer with commands to complete certain tasks. GAN gives machine learning a new power where it can skip human intervention and self-learn, maybe even imagine and create images that never existed.
In a general adversarial network, two neural networks are pitted against each other. These networks are provided with several images and the two networks, using data from the images creates new plausible images which are so real-looking, you could hardly tell in the first go. They learn from the data provided from several images such as facial expressions, complexions, bone structure, etc.
The neural networks require high end systems with huge data space to create the anticipated outcomes.
These networks trace the movements of pedestrians walking around which has helped develop driver-less cars. They have also studied voice modulations which make Alexa and Siri more conversational.
Deepfakes are using these deep learning techniques and hence the name.
Startups cashing in on deepfake technology
Synthesia is a London-based AI video generating platform where you can create corporate training videos for businesses around the world. The platform let’s you choose an avatar for the video and can be created in over 40 plus languages.
In grave times of the pandemic, it has made shoots riskier and Synthesia is slamming the corporate training costs by using deepfake technologies to create these videos.
Recently, it helped David Beckham appear in a public service announcement on malaria where he “spoke” in nine different languages.
Rosebud, founded by a team of AI experts, is aiming to disrupting media by creating glossy images used by e-commerce platforms or marketing. It has released around 25000 modelling images of people that don’t exist. It has also launched a service where you can put on clothes photographed on mannequins onto virtual real looking models.
This could be used by several cosmetic and apparel businesses who are constantly in requirement of models and who spend a fortune going through sets and shoots.
Modulate is using the deepfake technology in the gaming industry to analyze speech software that learns how to create Voice skins or filters of famous celebrities or characters. Old school method relies on text-to-speech tech whereas Modulate uses real time audio to synthesize the voice skins.
The voice skins have a real life effect which is radical for the gaming industry. Modulate.ai has raised $2 million through investors like Hyper plane venture Capital and 2Enable Partners.
Like Modulate, Respeecher demonstrates emotional nuance and prides itself in creating a technology of voice cloning. Respeecher aims at revolutionizing the film and animation industry. Since animation and film studios are prone to human error, Respeecher’s voice clones come in handy when actors who are dubbing, quit. Only a good sample of their voice and the technology does its wonders.
In the wake of privacy concerns with facial recognition synched with sensitive information, D-ID de-identifies and removes identifiable features without losing a grip on important attributes such as age, gender or emotion. Its smart anonymity feature provides top notch security for biometrics databases.
What are Deepfakes and how are they created?
Deepfake technology refers to audiovisual content generated by deep learning AI systems.
How do Deepfakes work?
Deepfakes rely on artificial neural networks, which are computer systems that recognize patterns in data.
What was the first Deepfake?
The first deepfake was coined on 2017 by a Reddit user of the same name. This user created a space on the online news and aggregation site, where they shared pornographic videos that used open source face-swapping technology.
The companies mentioned above are using deepfake tech to do more than just creating videos and images. They’re developing services around the most controversial phenomena and are being applauded for the same.
Lately, Deepfakes have become a common tool for harassment, revenge porn, busting political propagandas and are seldom used as a tool for harmless humor. There are several agencies relentlessly working on confirming deepfakes. But it has worried the entire AI tech industry since the technology is improving at lightning speed and are afraid that it might outrun the ones trying to find a difference or call out a fake.
Must have tools for startups - Recommended by StartupTalky
Subscribe to StartupTalky
Get the latest insights delivered to you right in your inbox