Leanne Manas is a familiar face on South African televisions. Towards the end of 2023 the morning news presenter\u2019s face showed up somewhere else: in\u00a0bogus news stories and fake advertisements\u00a0in which \u201cshe\u201d appeared to promote various products or get-rich-quick schemes.<\/p>\n\n\n\n
It quickly emerged that Manas had fallen victim to \u201cdeepfaking\u201d. Deepfakes\u00a0involve the use\u00a0of artificial intelligence tools to manipulate images, video and audio. And it doesn\u2019t require cutting-edge technical know-how. Software like\u00a0FaceSwap and ZaoApp, which can be downloaded for free, mean that anybody can create deepfakes.<\/p>\n\n\n\n
Deepfakes were initially used in the entertainment industry. For example, an actress in France who was unable to film her parts in person for a soap opera due to COVID restrictions still played the role\u00a0thanks to deepfakes. In the health industry deep-learning algorithms, which are responsible for deepfakes, are used to\u00a0detect tumours through pattern-matching in images.<\/p>\n\n\n\n
But these positive applications are few and far between. There are rising\u00a0global concerns\u00a0about the effect deepfakes might have on\u00a0democratic elections. Recent reports suggest that deepfakes are\u00a0on the rise\u00a0in South Africa and that\u00a0South Africans seemingly struggle to spot them.<\/p>\n\n\n\n
It is worrying, then, that South Africa\u2019s government hasn\u2019t yet taken any legislative steps to combat deepfakes \u2013 especially with the country\u2019s national elections scheduled for later this year. I am a legal scholar specialising in sport law, with a particular focus on image rights. <\/p>\n\n\n\n
I\u2019m especially interested in the recognition of an individual\u2019s image right and the legal position when their likeness is misappropriated without their consent. That includes the use of deepfakes.<\/p>\n\n\n\n
In my\u00a0LLD thesis, I argued that a person\u2019s image needs clear legal protection, taking into account the realities of digital media and the fact that many individuals such as influencers, athletes and celebrities generate an income from commodifying their image online. Promulgating legislation will create legal certainty in South Africa as it pertains to an individual\u2019s image.<\/p>\n\n\n\n
International examples<\/strong><\/p>\n\n\n\n Various states in the US have already taken action to deal with deepfakes, mostly in the context of elections. For example, Texas become one of the first states to\u00a0criminalise the use of deepfakes, especially if the content of the deepfake relates to political elections. It also recently passed\u00a0a second bill\u00a0which targets sexually explicit deepfakes. <\/p>\n\n\n\n So it\u2019s a criminal offence to create a deepfake video with the intention of injuring a political candidate or influencing an election result, or to distribute sexually explicit deepfakes without the consent of the individual, with the intention to embarrass them.<\/p>\n\n\n\n Maryland and Massachusetts, meanwhile, have proposed legislation that specifically prohibits the use of deepfakes. Maryland plans to target deepfakes that may influence politics; Massachusetts wants to criminalise the use of deepfakes for already \u201ccriminal or tortious (wrongful) conduct\u201d.<\/p>\n\n\n\n In 2020 California became the first US state to criminalise the use of deepfakes in political campaign promotion and advertising. The\u00a0AB 730 bill\u00a0makes it a crime to publish audio, imagery or a video that gives a false and damaging impression of a politician\u2019s words or actions. Though the bill doesn\u2019t explicitly mention deepfakes it is clear that AI-manufactured fakes are its primary concern.<\/p>\n\n\n\n In 2023, the governor of New York signed the Senate Bill 1042A. This aims to prohibit the dissemination of deepfakes in general, not just in relation to elections.<\/p>\n\n\n\n At least four federal deepfakes bills have been considered.\u00a0These include\u00a0the Identifying Outputs of Generative Adversarial Networks Act and the Deepfakes Accountability Act.<\/p>\n\n\n\n Protecting image rights<\/strong><\/p>\n\n\n\n There is currently no recognition of image rights in South Africa\u2019s case law or legislation. Image rights are distinct from copyright in law. The scope of protection provided by copyright alone would not be enough to tackle the problem of deepfakes in a court setting.<\/p>\n\n\n\n I\u00a0argue\u00a0for legal intervention which recognises individual image rights. By recognising an image right the image will be protected against unauthorised use. <\/p>\n\n\n\n This will not only include the misappropriation of an individual image for commercial use, it will also combat deepfakes, whether those relate to elections and politicians or any manipulation of a person\u2019s image with malicious intent.<\/p>\n\n\n\n Image rights legislation is key. It can:<\/p>\n\n\n\n This can all help regulate deepfake situations. The malicious and deceptive nature of deepfakes may cause the image right holder to suffer significant harm. It is time that South Africa\u2019s legislature addressed these situations by providing the necessary protection to individuals.<\/p>\n\n\n\n\n