Nobody can tell which actress is real in Scarlett Johansson deep fake challenge

Deepfake technology is growing increasingly sophisticated, and with it, the potential for members of the public to be fooled by fake videos.

The way deepfakes have been used so far ranges from the shocking to downright dangerous. Deepfakes use artificial intelligence to create convincing fake images or videos of real people, in order to make them say or do things that the real person never did.

Deepfakes of politicians like Joe Biden have led to misinformation on social media, while the technology has also been used to superimpose celebrities and particularly women in illegal explicit videos.

READ NEXT: Queen says she loves 'Netflix and Phil' in Channel 4 'deepfake' Christmas message

The terrifying potential of deepfakes has been demonstrated again with a side-by-side video of Scarlett Johansson and Elizabeth Olsen, with viewers asked to guess which is a deepfake and which is real.

The clip shows the two actresses talking in a video interview, saying the same words and using the same facial expressions as each other in an identical way.

Although both clips look realistic, eagle-eyed Twitter users were able to spot which video was a deepfake thanks to some inconsistencies in the footage.

One user pointed out that the clip of Johansson shows her with a 'double eyebrow'.

Other telltale signs of a deepfake in the clip include the lack of shadows on Johansson's face when she moves her hand, flaws in the lighting on her face, and peculiarities in her hair and neck muscles.

  • Pogba and Neymar skins for Modern Warfare II leak with Messi tipped to follow

One viewer tweeted: "This is really scary" while another wrote: "Still wild how close they look!".

Public awareness of deepfakes has been growing for some time, amid fears that misinformation online will only grow in scale.

In 2020, a deepfake image of Tom Hanks was created for less than £100 to demonstrate how easily the AI doppelgangers can be created.

The image's creator, Philip Tully, said: "People with not a lot of experience can take these machine-learning models and do pretty powerful things with them."

READ MORE:

  • Call of Duty fans go wild for Modern Warfare II graphics that 'look like real life'
  • iPhone chargers will officially stop working on new models and be replaced with USB-C
  • Killer autonomous robot army deployed by major NATO country in terrifying world first
  • McDonalds sued for £805million by ice cream hackers who tried to fix McFlurrys
  • 'End of an era' as iconic millennial games site Miniclip closes down for good

Source: Read Full Article