Authored By: Zulaikha Khan
STADIO Higher Education
Introduction
A case in point occurred in March 2024 when three South African investors were put out of pocket after falling for social media ads. The advertisements featured what appeared to be the entrepreneur Elon Musk and businessman Johann Rupert vouching for a trading platform, even though neither had any part in it. The video was a fabrication of generative artificial intelligence, a tool that can now put real or made-up words into a person’s mouth with an uncanny degree of visual accuracy.
South Africa has been quick to legislate against one form of this abuse. Under section 16 of the Cybercrimes Act 19 of 2020, the non-consensual release of an intimate image is a criminal matter; the Act’s definition is broad enough to cover both genuine and simulated pictures, making sexual deepfakes unlawful.[1] But there is no such statutory provision for non-sexual deepfakes, those pieces of footage which may humiliate or degrade an identifiable individual without resorting to nudity or sexual conduct. This article puts forward the view that the common-law offence of crimen injuria could be applied to such behaviour on doctrinal grounds, as its focus on the impairment of dignity is not contingent on the veracity of the depiction. Yet this is an untested area before the courts and legislative action would be the surest way to plug the hole. What follows is an examination of the current legal position, where it falls short, and a proposal for reform.
The Existing Legal Framework
One can trace two lines of authority in South African law when it comes to fabricated digital content: statute and the common law. Part II of the Cybercrimes Act deals with malicious communications.[2] It is an offence under section 16 to use an electronic service to put out a data message with an intimate image of another without consent.[3] The Act is clear that “intimate image” includes a simulation of a person in a state of nudity or sexual conduct, so a pornographic image created by AI is covered even if the subject never did anything of the sort.[4] There is also the Films and Publications Act 65 of 1996, which offers a more limited shield against the unauthorised distribution of private sexual material.[5] In either instance, the provisions are written to apply to sexual or intimate matters only.
For harm that is humiliating but not sexual in nature – say, a doctored video of someone making racist comments or admitting to a crime – the recourse is the common-law crime of crimen injuria. Put simply, it is the intentional and unlawful violation of another’s dignity or privacy, provided the transgression is not trivial.[6] The Mostert case shows how easily the offence can encompass direct verbal abuse of a degrading kind. Precedent has also been set for indecent exposure and invasions of privacy, including peeping-Tom type behaviour, where the perpetrator is dealing directly with a complainant who is present.[7] To date, however, reported authority does not show the crime being used against a third party who has manufactured and put around a wholly invented depiction without ever coming into contact with the victim.
The Gap: Fabrication Without Sexual Content
The issue is whether the dignity-impairment at the heart of crimen injuria must be grounded in some truth about the complainant. For comparison, one might look at criminal defamation, which the Supreme Court of Appeal in S v Hoho has upheld as constitutionally sound.[8] Defamation is predicated on the publication of something that will damage a reputation, and the doctrine has always treated the offending matter as a factual claim. If an audience is taken in by a convincing deepfake, their belief in the matter injures the complainant’s standing and criminal defamation provides a remedy.
But there is a gap where the deepfake is transparently fake yet still degrading. A viewer in the know is not deceived, so traditional defamation has no purchase on the situation. It is this narrower category of conduct that we address here.
The structure of crimen injuria is not the same. What it is concerned with is the dignity and privacy of the complainant, as opposed to an audience’s credulity in a false claim. One can submit that on this basis a made-up depiction is capable of impairing dignity even if its falsity is plain to see; the injury is in being made an unwilling object of degrading material, not in whether one has been taken in by third parties. A person who is shown in a synthetic video doing something they never did endures a serious and real affront to their dignity, viewer belief in the footage or not.
There are difficulties with such a reading. It is true that the unlawfulness element of crimen injuria has in the past put true statements and fair comment out of bounds, but no South African court has yet ruled on whether a defence of obvious fabrication would be available for a deepfake that is not deceiving anyone. Nor has there been any consideration given to whether animus iniuriandi is present when the creator’s aim was satire or political commentary and not a deliberate intent to humiliate.[9] The result is that while one can make a doctrinal case for applying crimen injuria to non-sexual deepfakes, it is an uncertain proposition and an accused might well try to draw an analogy with the truth-based logic of defamation.
III. Towards Reform
To put an end to this uncertainty there are two options. A court could take the judicial route and decide in a fit case that the offence’s definition of dignity holds regardless of veracity, thus extending its implicit reasoning to synthetic media. But this is subject to the usual perils of common-law evolution via litigation: it is slow, can be inconsistent and is contingent on a proper case making its way to a superior court.
Legislation is the surer course. Parliament has already plugged the hole left by sexual deepfakes under the Cybercrimes Act by broadening the scope of intimate images to include simulations. An amendment along those lines could bring the Act’s malicious communications provisions to bear on any fabricated audio-visual of an identifiable person put out without consent to harass or degrade, sexual or otherwise. This would do away with the need to rely on an untested stretch of the common law while keeping in place the Act’s procedural protections for complainants.[10]
Such a legislative fix would also honour the constitutional equilibrium between expression and dignity more effectively than an uneasy common-law extension.[11] Parliament is in a better position to set down express exceptions for parody and legitimate commentary than criminal courts are in the course of adjudicating individual cases. As Gotora has put it, the current regulatory framework in South Africa was not conceived for generative synthetic media and calls for reform of a kind that cannot be muddled through with older doctrines.[12]
Conclusion
In dealing with the harm of sexual and intimate deepfakes, South African law has been decisive. Yet where non-sexual content is fabricated to cause humiliation or reputational damage, there is no comparable statutory response. Crimen injuria is a plausible way to bridge that divide given its focus on impaired dignity over falsehood, but with no reported decision on the matter to guide them, victims and prosecutors are left in the dark. This article has made the case for amending the Cybercrimes Act 19 of 2020 so that its provisions on malicious communications cover non-sexual depictions of identifiable persons in the manner already applied to simulated intimate images. Absent that, a complainant faced with a video or image that is mocking or degrading will have to make do with a corner of the common law that was not meant for, and has not been put to the test against, this type of harm.
Table of Cases
S v Hoho 2009 (1) SACR 276 (SCA).
S v Mostert 2006 (1) SACR 560 (N).
Table of Legislation
Constitution of the Republic of South Africa, 1996.
Cybercrimes Act 19 of 2020.
Films and Publications Act 65 of 1996.
Films and Publications Amendment Act 11 of 2019.
Bibliography
Burchell J, ‘Protecting Dignity Under Common Law and the Constitution: The Significance of Crimen Iniuria in South African Criminal Law’ (2014) 27(3) South African Journal of Criminal Justice 250.
Gotora NT, ‘Unmasking Deception: Deepfake Regulation in the Context of South African Law, Could a Rethinking of Performers’ Protection Rights Be the Answer?’ (2024) 32 International Journal of Law and Information Technology eaae026.
Snyman CR, Criminal Law (4th edn, Butterworths 2002).
[1]Cybercrimes Act 19 of 2020, s 1 (definition of ‘intimate image’); s 16.
[2]ibid, Part II.
[3]ibid, s 16(1).
[4]ibid, s 1 (definition of ‘intimate image’).
[5]Films and Publications Act 65 of 1996, s 18F (as inserted by the Films and Publications Amendment Act 11 of 2019).
[6]C R Snyman, Criminal Law (4th edn, Butterworths 2002) 453; S v Mostert 2006 (1) SACR 560 (N) 571b-c (Theron J).
[7]Jonathan Burchell, ‘Protecting Dignity Under Common Law and the Constitution: The Significance of Crimen Iniuria in South African Criminal Law’ (2014) 27(3) South African Journal of Criminal Justice 250, 251.
[8]S v Hoho 2009 (1) SACR 276 (SCA) para 23.
[9]Snyman (n 7).
[10]Cybercrimes Act 19 of 2020, Part VI, ss 20-22.
[11]Constitution of the Republic of South Africa, 1996, ss 10, 16.
[12]Gotora (n 1).
