Business
Deepfake of Elon Musk kidnapped used in marketing campaign
Close
A marketing deepfake video of Elon Musk has gone viral online, gaining nearly 1,000,000 views in just a few days.
The clip shows Elon held hostage in a warehouse by real-estate investment startup reAlpha and explains the companies mission along with some satirical jokes. The quality of the deepfake isn’t very convincing but the voice sounds fairly accurate.
The video acknowledges that this isn’t the real Elon Musk and that the parody was created as under Fair Use as entertainment and education on short term rentals and fractional real estate model and that no malice is intended.
Who created the deepfake?
reAplha
Nate Doane
Slack Shack Films
Was the content disclosed as a deepfake?
Yes
Was the deepfake consensual?
No
How was the deepfake created?
Video Editing
Deepfake Software
Video Production Team
Year :
2022
Indian superstar promotes local stores using deepfake technology
Close
In support of small businesses that suffered during covid, Cadbury helped local owners by using India's biggest brand ambassador, Shah Rukh Khan, alongside deep machine learning to show the actor saying names of local stores and brands.
Deepfakes were used to recreate Khan's face and voice to take the local store names in the ads. Different versions of the same ad with local store names were targeted as per the pin code of the viewer, showing them only nearby stores. People were able to create their own version of the ad, allowing any small business owner to promote their stores on social media pages and apps such as WhatsApp.
During Diwali, Not Just A Cadbury Ad gave 'power to the people' to create their own business with the Bollywood star turning into a personal brand ambassador for more than 85,000 struggling retailers using deepfake technology to help local business grow. The campaign yielded unprecedented results such as delivering 22% higher VTRs and 18% higher CTRs. More than 105,000 users logged on to the website to create their own version of the ad. A total of 139k versions of the ad were created which were circulated through WhatsApp and other social media. The personalised campaign won a Titanium Lion award at Cannes Festival.
Who created the deepfake?
Cadbury
Rephrase.ai
Wavemaker
Ogilvy
Was the content disclosed as a deepfake?
Yes
Was the deepfake consensual?
Yes
How was the deepfake created?
Deepfake Software
Video Editing
Rephrase.ai
Wavemaker
Year :
2021
Deepfake technology allows anyone to dance like a pro
Researches created a program that copies dance moves of one person and 'gifts' them to another by making it look as if they can dance using neural-networking software.
It took several months for the team to map the face of one person to the entire body of another. The video shows the original source performing and then various other shots of people never actually doing what they have done but performing perfectly thanks to deepfake technology. The process captures two neural (generative adversarial) networks, videos of one person dancing and another video of someone moving. The first movement is mimicked by the second dancer and additional software is then used to refine the routine and ensure the face appears clear.
The UC Berkley researchers created a paper on "do as I do" motion transfer. They say the project 'hit a nerve with folks', especially disabled people, for example quadriplegics, who wanted to visually watch themselves move in ways they were unable to do physically. Due to a simple method producing compelling results, the researchers were motivated to provide a forensics tool for reliable synthetic content detection. Additionally they also released an open-source dataset of videos available to the public interested in motion transfer.
Who created the deepfake?
UC Berkeley researchers
Caroline Chan
Shiry Ginosar
Tinghui Zhou
Alexei Efros
Was the content disclosed as a deepfake?
Yes
Was the deepfake consensual?
Yes
How was the deepfake created?
Deepfake Software
Video Editing
File Footage
Motion transfer
Target subject performing
Spatio-temporal smoothing
Realistic face synthesis
Year :
2018
China reveals first AI news anchor
Xinhua, China’s state news agency introduced AI anchors who will report “tirelessly” all day every day, from anywhere in the country.
A digital version of the news anchor dons a pin-striped suit with a red tie and realistically nods his head, blinks and raises his eyebrows. He boasts 24 hour news coverage every day of the year so he can bring the news to viewers no matter the scene. Another AI anchor was created to speak English and they both acknowledge they are wholly cloned and can provide the news once they are provided with text. Both were built using machine learning to simulate audio, facial expressions, and gestures of real-life people.
According to Xinhua, the AI technology has endless capabilities with examples of popular book reading apps telling stories from family members themselves one day.
Vision of the Chinese AI anchor quickly spread on social media with many viewers as impressed as they were alarmed. “A little bit horrible,” one said, to which another responded: “Really scary.” Xinhua are experimenting with the software as a way to lower production costs.
Who created the deepfake?
Xinhua
Sogou
Was the content disclosed as a deepfake?
Yes
Was the deepfake consensual?
Yes
How was the deepfake created?
Deepfake Software
Video Editing
Facial Recognition
Facial Modeling
Speech Synthesis Software
Year :
2018
Journalist deepfaked committing a crime he didn't do
There’s a video of Michael Grothaus attempting an armed robbery but the author explains this never happened and is in fact a deepfake that he commissioned.
The video shows the victim, a cyclist, whose head-mounted GoPro camera recorded the ordeal. Suddenly the clip shows Grothaus riding past on a motorcycle, pulling out a gun and forcing him to stop in his tracks and give him his rucksack, but the cyclist says he doesn’t understand Spanish, a language Grothaus can't speak in real life, but in the deepfake he masters it. He jumps off the bike and Grothaus chases him by foot with anxiety and fear on his face but the chase is stopped by a bystander who intervenes.
Grothaus had the fortune of knowing the deepfake was coming. He commissioned it whilst researching his latest book, Trust No One: Inside the World of Deepfakes. He told the deepfaker he wanted to ‘commit’ a crime but knew nothing else so the element of surprise would be similar to that of an innocent civilian. He explains that anyone can be the next victim of deepfakery and that every second of video is made up of about 30 photos, which means for every minute of video posted to social media, the deepfaker has 1,800 photos of your face.
Who created the deepfake?
Unknown Deepfake Artist that was Commissioned
Was the content disclosed as a deepfake?
Yes
Was the deepfake consensual?
Yes
How was the deepfake created?
Deepfake Software
Video Editing
Existing footage provided from the internet
Year :
2021
China unveils world's first female AI news anchor
China’s state-run news agency, Xinhua released their latest news anchor Xin Xiaomeng, a computer generated model powered by advanced AI technology.
The AI news anchor perfectly delivers her message in Chinese, never missing a cue or stuttering and is based off human newsbroadcaster, Qu Meng. In the video she announces her professional debut will be in March. It’s unknown if she’ll speak in English too.
Who created the deepfake?
Xinhua
Sogou
Was the content disclosed as a deepfake?
Yes
Was the deepfake consensual?
Yes
How was the deepfake created?
Artificial Intelligence
Deepfake Software
Video Editing
Year :
2019
New tool generates deepfake videos from one single image
Close
MegaPortraits can overlay still photos of people with a realistic and expressive moving version of another person.
Samsung Labs built a new tool that generates realistic deepfakes with just a single photo or painting that can imitate facial expressions, head and neck stretching, and the other subtle elements that make a human face come to life.
The researchers unveiled a novel method for creating high-resolution neural avatars. The AI model generating the movement is trained from two random sample frames from the source and driver at each step. The high-resolution avatars are designed to function regardless of whether someone looks like the image they are imitating. The group’s efforts also include working on creating the deepfakes in real-time. The AI model connects 100 neural avatars linked to existing images and runs at 130 frames per second.
With simple deepfake tools being easily made to the public, there are concerns of non-consensual content, especially when it comes to celebrities and resurrecting the deceased.
Year :
2022
Hackers use deepfake of Binance CCO in exchange listing scams
Close
In a series of deepfake video calls, hackers have impersonated Binance chief communications officer, Patrick Hillmann, fooling several representatives of cryptocurrency projects.
The attackers used what Hillman described as an AI hologram deepfake of his image, tricking some into thinking he was helping them list on the Stock Exchange.
The COO says there’s been a recent spike in hackers pretending to be Binance employees and executives on social media platforms such as Twitter, LinkedIn and Telegram but says pre-bunking helps to be prepared, as well as defend users and the ecosystem.
Who created the deepfake?
Unknown
Was the content disclosed as a deepfake?
No
Was the deepfake consensual?
No
How was the deepfake created?
AI hologram deepfake
Year :
2022
Bloomberg reporter's deepfake avatar introduces news in English and Spanish
Close
Bloomberg has aired Tom Mackenzie’s avatar to introduce the news segment in English and fluent Spanish, a language he is yet to learn.
Sitting behind a breaking news screen, the realistic avatar welcomes his audience to the European market open on Friday morning. The only thing giving away Mackenzie's digital anchor is his co-host giggling about his voice, which was described as “smoother” than usual.
The segment explores the future of synthetic media, while showcasing Mackenzie’s English and Spanish-speaking skills.
Who created the deepfake?
Synthesia
Bloomberg
Was the content disclosed as a deepfake?
Yes
Was the deepfake consensual?
Yes
How was the deepfake created?
Synthesia
Video Editing
Year :
2022
Bruce Willis promotes Russian mobile network in Die Hard deepfake
Close
Bruce Willis has made a new appearance as John McClane, only this time, deepfake technology has made the process much easier.
The Russian phone ad series shows not Bruce Willis himself, but an authorised impersonator reprise the role from action thriller film Die Hard. Deepcake used an artificial neural network to impose Bruce Willis' image onto the face of a Russian actor of similar age and appearance to create a 'deepfaked digital twin'.
Deepcake CEO said the process of teaching neurone algorithm is a legnthy one. The first training video took up to 17 days to make. After that the process was sped up to produce footage up to 4k resolution. The company is now working on generating faces in up to five days.
On-screen, the Hollywood star perfectly masters the Russian language thanks to synthetic media. Off-screen, Willis’ consent to the deepfake ad means he is able to reap the benefits of $1-2 million from the comfort of his home in the US - majorly cutting down production costs. News of Bruce Willis becoming 'the first Hollywood star to sell his rights to allow a digital twin of himself to be created for use on screen' spread like wildfire but turned out to be incorrect. The allegations come after the actor, who's stared in more than 70 films, was diagnosed with aphasia which causes difficulty with speech. Willis believes the advance of new technologies is an interesting experience and allows participation from across the globe after the success of his deepfake ad campaign.
Who created the deepfake?
Deepcake
MegaFon
Was the content disclosed as a deepfake?
Yes
Was the deepfake consensual?
Yes
How was the deepfake created?
Deepfake Software
Video Editing
Artificial neural network
Year :
2021
Snoop Dogg features as a deepfake in Australian TV commercial
Close
When you hear ‘did somebody say,’ you instantly think ‘Just Eat,’ thanks to the well-known ad sung by Snoop Dogg.
JustEat’s Australian subsidiary, MenuLog, wanted to leverage a similarly successful campaign without spending large sums of money or using major production crews multiple times.
Using footage from the original ad, along with Snoop Dogg’s consent, Synthesia’s deep learning algorithms repurposed the original video to a localised version. It turns out the use of synthetic media was so good that it fooled Australians who found it bizarre that a 'paid ad' disclaimer was included on the ad. Not many could tell that it wasn't Snoop but instead his synthetic avatar singing on screen.
Who created the deepfake?
Synthesia
MenuLog / JustEat
Was the content disclosed as a deepfake?
Yes
Was the deepfake consensual?
Yes
How was the deepfake created?
Synthesia
Video Editing
Year :
2020
Lionel Messi personalises video messages for all
Close
Messi Messages is a simple video experience enabling fans to send personalized videos with Messi addressing their friends by name.
To create Messi’s synthetic avatar, Synthesia recorded the soccer superstar for five minutes in a green screen studio, speaking his native language - Spanish. To make it a more global experience, his synthetic avatar also learnt how to mimic eight different languages.
Within 24 hours the campaign had 38 million hits. Messi himself shared the experience with his 192+ million Instagram followers, triggering countless Messi Messages to pop up across every social media platform and in group chats across the world – with many double-takes in the process.
Who created the deepfake?
PepsiCo brand Lay’s
UNIT9
180Amsterdam
Synthesia
Was the content disclosed as a deepfake?
Yes
Was the deepfake consensual?
Yes
How was the deepfake created?
Synthesia
Deep-learning algorithms
Year :
2022
Deepfake of ESPN reporter predicting future Chicago Bulls' wins in 1998 TV ad
A TV commercial has caused widespread debate after it showed footage of an ESPN analyst from 1998 making disturbingly accurate predictions for 2020.
During The Last Dance documentary, State Farm’s ad showed a much younger Kenny Mayne reporting from the sports desk predicting that in 20+ years there would be a series created about the Chicago Bulls, tight jeans would come into fashion and he also uses the word ‘lit’.
As filming commercials were hindered due to Covid-19, ad agencies turned to technology to create content. Mayne’s hunch confused several viewers and the ad was widely discussed.
Who created the deepfake?
State Farm
Was the content disclosed as a deepfake?
No
Was the deepfake consensual?
Yes
How was the deepfake created?
Deepfake Software
Video Editing
Visual Effects
Archival footage
Facial Mapping Technology
Year :
2020
AI-generated Elon Musk joins Zoom call
Deepfake Elon Musk unexpectedly appeared on a video call in real-time with two very surprised professionals.
The fake Elon Musk also known as programmer Ali Aliev, used an open source code from face-swapping software. Reportedly Aliev took only a couple of hours to develop the prototype which the team saw huge potential in and built on.
Avatarify is now an available open-source program that deepfakes people during video conferencing calls in real-time. Filters include the likeness of celebrities, but the low quality of the face swap is evident. The code is available on Github for anyone to use.
Who created the deepfake?
Ali Aliev
Was the content disclosed as a deepfake?
Yes
Was the deepfake consensual?
No
How was the deepfake created?
Deepfake Software
Video Editing
Visual Effects
First Order Motion
Year :
2020
Deepfake artwork allows for actress’ time travel
Close
AI has enabled an artist to bring Italian actress, Ornella Muti, 'back to the stage'.
Joseph Ayerle’s deepfake artwork shows Muti’s face from 1978, superimposed onto the body of Kendall Jenner and was created without a Hollywood budget.
Who created the deepfake?
Joseph Ayerle
Was the content disclosed as a deepfake?
No
Was the deepfake consensual?
Unknown
How was the deepfake created?
Deepfake Software
AI Neural Network
Video Editing
Kendall Jenner as a 'body double'
2000 Picture Analysis of Ornella Mutis Face
Year :
2018
.gif)
Mona Lisa moves for the firs time in deepfake
An AI-generated living portrait of Leonardo da Vinci’s Mona Lisa went viral as people saw the iconic lady move her lips, nod her head and look from side to side.
Researchers in Moscow developed the videos from just one single image. Using the method known as single-shot learning, researchers also brought celebrities back to life, including Marilyn Monroe and Albert Einstein.
Who created the deepfake?
Samsung’s AI lab
Skolkovo Institute of Science and Technology
Was the content disclosed as a deepfake?
No
Was the deepfake consensual?
Unknown
How was the deepfake created?
Deepfake Software
Few-Shot Adversarial Learning of Realistic Neural Talking Head Models
Year :
2019