JOHAN STEYN: The devil’s in the deepfake

The “post-truth era” is filled with altered images indistinguishable from real ones, used for manipulation and clickbait.

Charming as always, Elvis recently appeared on the popular TV show America’s Got Talent. Performing his 1968 hit You’re the Devil in Disguise, he was joined on stage by judges Simon Cowell, Heidi Klum and Sofia Vergara. 

“You look like an angel, walk like an angel, talk like an angel, but I got wise: You’re the devil in disguise.”

A start-up called Metaphysic has managed to advance to the talent competition’s final round. Garnering attention for its hyper-realistic deepfakes, Metaphysic is generating revenue by crafting deepfakes for use in the film industry and marketing sectors.

The company has launched an ancillary venture, dubbed Every Anyone, which offers individuals the ability to create convincing deepfakes simply by uploading a selfie.

Deepfake involves digital manipulation using AI and machine learning to create or alter video and audio content while maintaining that of being real. The technology enables images and videos to be superimposed onto others to create an image or video that appears unaltered. 

Deepfakes can convincingly depict people saying or doing things they never did, often making it challenging to distinguish between genuine and fabricated media. While the technology has creative applications, such as in entertainment, it also raises important ethical, legal, and security concerns due to its potential for misuse.

Scams using deepfakes of well-known individuals have emerged, including a fake video of SABC news anchor Francis Herd promoting a fraudulent investment scheme linked to Elon Musk. The video advertises an attractive return on investment and features an AI-generated version of Herd, followed by a deepfake Musk endorsing a supposedly revolutionary investment software.

The video has circulated on YouTube and Facebook, with supposed endorsements from Nicky Oppenheimer and Johann Rupert.

Leanne Manas, the host of SABC’s Morning Live, has become a victim of online scam operations using AI technology, leveraging her reputation by attaching her name and altered images to clickbait material. They include doctored photos placed alongside fabricated news articles and advertisements implying her endorsement. 

Despite her efforts to solicit social media platforms to crack down on these abuses, the response has been minimal. Manas has been warning the public no to click on such deceptive links after a number of people have been duped by these scams, particularly those promoting weight loss (“Keto Gummies”) and get-rich-quick schemes.

The misuse of deepfake technology has taken a particularly disturbing turn with its application in the creation of non-consensual pornography.

A staggering 96% of deepfake content falls into this category, according to recent statistics, underscoring the grim reality where women are exploited without their permission.

These deepfakes are often indistinguishable from real videos, which can cause irreparable damage to the reputations and lives of those depicted. 

Activists and lawmakers are beginning to address these issues, calling for stringent measures to curb the creation and distribution of deepfake pornography. However, the fight against this misuse is complex; it involves not only legal challenges but also the need for technological interventions to detect and prevent the spread of these falsified images.

Humans are storytellers — but what happens when we can no longer believe stories? The “post-truth” era is in full swing with the help of AI, the devil in disguise.

• Steyn is a human-centred AI advocate and thought leader. He is the founder of AIforBusiness.net.