These Realistic Deepfakes of Sly Stallone as The Terminator are Terrifying

A new YouTube channel known as Ctrl Shift Face is quickly gaining attention online with a series of videos featuring scenes from popular movies re-imagined by superimposing actors' faces onto others.

The channel's latest pair of videos, playing with the concept of an alternate reality where Sylvester Stallone took the starring role in the 1991 Hollywood blockbuster Terminator 2, has left viewers amazed, terrified, and wanting more.

"That feel when some YouTube guy's CGI is better than Hollywood Terminator Genisys face replacement CGI," one user wrote. "The shadows. How the bleep did you pull off those shadows?" another stunned viewer asked. "I wish I could get stock in this channel before it explodes," a third chimed in.

Others took a more serious turn, discussing the technology's potential real-world applications. "This is scary. The legal implications are gonna be unmanageable in the future. The end is definitely near," one concerned viewer wrote. "Jesus Christ, this technology is terrifying," another added. "I'll keep enjoying this YouTube channel until you get recruited full time by the FBI," one user joked.

Ctrl Shift Face began gaining attention last month when he uploaded a video featuring former Saturday Night Live player Bill Hader impersonating Arnold Schwarzenegger, with the clip seamlessly cutting to superimpose the Hollywood Star's face onto Hader's. That video has garnered nearly five million views.

Deepfakes make use of an artificial intelligence learning technique known as 'generative adversarial network', with artificial intelligence trained to detect the facial expressions and mannerisms of celebrities, politicians and other public figures to mimic them in doctored clips through face-swapping technology based on publicly available film and television appearances.

The emerging technology, popularised in late 2017, has raised grave concerns over the potential to use deep fakes to create fake news, fake celebrity pornographic videos, and other malicious content, particularly as the technology becomes more and more impressive, available, and difficult to detect.