A recent controversial video showcasing an actress resembling Wonder Woman star Gal Gadot in an adult film has raised concerns about the misuse of machine learning technology.
The video, crafted by a Reddit user known as deepfakes, portrays a woman resembling Gadot, with the actress’s face digitally imposed onto someone else’s body.
This manipulated footage was generated through the training of a machine learning program on various images and videos of Gadot sourced from stock photos, Google searches, and YouTube clips. This concerning development highlights how easily accessible and advanced this technology has become.
Watch the video below for more details.
A concerning new video depicting Gal Gadot in a short adult film has brought attention to the dangers of misuse of machine learning technology. While the video may not deceive many, it raises alarms about the potential for creating fake pornography and other harmful content using machine learning without consent. This is not an isolated incident, as other celebrities have also been targeted by similar manipulative techniques. The individual behind this video utilized open-source machine learning tools to superimpose Gadot’s face onto another person’s body. This process involved training an algorithm on existing adult content and images of Gadot to achieve a realistic facial overlay. The ease and speed at which these fake videos can be created using accessible resources is troubling and highlights the need for increased vigilance against such malicious activities.
The video was created using a machine learning algorithm trained on a mix of stock photos, Google images, and YouTube videos featuring the star (depicted in the image above as Wonder Woman) – a process experts now describe as straightforward. According to Motherboard, the constant stream of self-uploaded photos on social media platforms makes it possible for someone to use such a method for malicious purposes.
AI researcher Alex Champandard emphasized the importance for everyone to understand how easily images and videos can be manipulated, to the point where detecting counterfeits may become impossible in just a few months. Champandard pointed out that while this kind of manipulation was previously only achievable with significant resources and visual effects expertise, it can now be accomplished by a solo programmer equipped with modern computer hardware.