Sameera Casmod | sameerac@radioislam.co.za
29 February 2024 | 12:22 p.m. SAST
2 minute read

Image: IAS Exam

It’s been dubbed the 21^st century’s response to Photoshop—but on steroids. A product of deep learning and artificial intelligence, deepfakes involve the manipulation of images and videos to swap faces, voices, and even entire personas, and their risks extend far beyond those of Photoshop.

“They use an application where they swap your face, your voice and manipulate videos. So they essentially make a whole new image from images that are already available,” Dr Layckan van Gensen, a junior lecturer at Stellenbosch University, explains.

Although it has positive uses, such as its application in detecting tumours in patients, it also has many drawbacks. It has been used to dupe people into paying money, like the finance worker who paid $25 million to fraudsters after a video call with the supposed chief financial officer of the company.

It has been used extensively in the global political arena, especially during elections. An AI-synthesised impersonation of the voice of an opposition leader heavily swayed election results in Slovakia in October last year, according to a report. Deepfakes have been used to manipulate elections in Pakistan, Bangladesh, and the United States. The list and concerns grow in an election bumper year.

South Africa is witnessing the increasing use of deepfake technology, prompting calls for the establishment of legislation to protect a person’s image. With the proliferation of social media and easily accessible deepfake apps, individuals are increasingly vulnerable to exploitation and misinformation.

“Deepfakes are mostly used to distribute misinformation. And…it is an election year, and misinformation cannot be distributed because then we don’t know what is true and what is not,” Dr van Gensen notes.

During the legislative elections in India in November last year, a deepfake video was circulated on social media just before voting had started, telling voters to cast their ballots in favour of Congress, the country’s main opposition party, Dr van Gensen, an attorney at law, illustrates.

Dr van Gensen advocates for the government to establish legislation to protect image rights and, in turn, deter the use of deepfakes in South Africa.

However, legislation can only be enforced in the same jurisdiction as where the victim resides, posing enforcement challenges for perpetrators operating beyond the reach of local authorities. Dr Layckan suggests implementing standards for AI application access to mitigate the proliferation of deepfake technology.

Listen to the full interview on The Daily Roundup with Moulana Junaid Kharsany here.