On 8 October 2026 the South African Human Rights Commission (SAHRC) condemned the rape and death threats posted online against March and March leader Jacinta Ngobese-Zuma. TimesLIVE reports that she opened a criminal case in Durban against a purported ANC member over Facebook posts allegedly calling for her rape and murder. Those allegations have not been tested in court. The SAHRC said social media platforms are "responsible for filtering messages posted on their platforms" so that they do not amplify hate speech and gender-based violence. It also urged the executive to bring the Prevention and Combating of Hate Crimes and Hate Speech Act of 2023 into operation quickly. Its statement names no platform.
The two demands rest on different legal footing. Only one of them is something the Act can deliver.
The strongest case for the commission's position
The case for acting is serious. Threats of rape and murder against a woman who leads a public campaign are meant to drive her out of public life. The SAHRC reminded the public that threatening or calling for rape is a criminal offence and falls outside constitutional protection of expression. Platforms also run ranking and recommendation systems that can make a threat more visible. A regulator that asks them to take proportionate care is not asking for censorship. Victims who report threats to a platform and get silence or a canned reply have good reason to want a statutory duty.
I accept all of that. The question is whether the instruments being demanded fit the harm.
What the Act actually does
The Act was assented to on 6 May 2024 and published in Government Gazette 50652 on 14 May 2024. Section 12 says it "comes into operation on a date fixed by the President by proclamation in the Gazette." The commission's own call, and the lack of a proclamation reported to date, indicate that has not happened. The Parliamentary Monitoring Group's bill record shows the bill was introduced on 12 April 2018, lapsed in 2019, was revived that October, and passed both Houses on 5 December 2023.
Section 4(1)(a) creates the offence of hate speech. It applies to anyone who intentionally publishes, propagates, advocates, makes available or communicates something that could reasonably be construed as showing a clear intention to be harmful or to incite harm, or to promote or propagate hatred, based on listed grounds. Section 4(1)(b) extends this to a person who intentionally distributes or makes available an electronic communication that they know is hate speech, through an electronic communications system.
The Act therefore reaches the speaker, and anyone who knowingly distributes the speech. It contains no duty on platforms to filter, monitor or remove content. The Act's definitions section does define "electronic communications system", but only to describe the channel through which a person commits the offence. Penalties under section 6(3) are a fine, imprisonment of up to five years, or both. Section 4(3) requires the Director of Public Prosecutions to authorise any prosecution, and section 4(2) exempts good-faith artistic, academic, public-interest reporting and religious expression, provided it does not advocate hatred that incites harm.
Two points follow. First, commencing the Act would give prosecutors an additional charge against the person behind a post. It would not make Facebook or X filter anything. Second, the hate speech offence turns on "grounds" in section 1 such as gender, race and religion. A rape threat against a woman could plausibly engage the gender ground, but the SAHRC itself says threats of this kind are already criminal. Delay in commencement is not the obstacle to the case in Durban.
Why a filtering mandate is the weaker lever
The SAHRC's platform language is general, and it is worth being precise about the risks of a legal rule that matches it. A duty to filter messages means automated screening of every post. Such systems cannot judge intent, so they flag reclaimed slurs, news reports quoting a threat, and victims posting screenshots as evidence. The Act's own section 4(2) exemption for fair and accurate public-interest reporting depends on context and good faith, which a classifier does not see. Platforms facing liability for what they fail to catch would tend to over-remove, and the speech lost would disproportionately be that of the people the law is meant to protect.
There are narrower options. A rape or murder threat sent to a named individual is a clear, rare category that platforms can prioritise through human review without screening everything. Faster and standardised preservation of evidence would help police, who need account data and screenshots to identify and charge a suspect. Transparent reporting on how long platforms take to act on threat reports would let regulators and the public measure performance. These steps target the real failure points, which are slow response and weak cooperation with investigators, and they leave general-purpose speech alone.
What the executive should do
Commencement is the executive's decision. The Act has stood unproclaimed for roughly 29 months, and the President can fix a date by proclamation at any time. If the government believes the Act's text is ready, it should proclaim it and let prosecutors and courts test the section 4 offence, including its good-faith exemptions. If it has concerns about the text, it should say so and amend it. Leaving a signed statute in limbo gives neither victims nor speakers any certainty.
The commission should also say what it wants platforms to do, in terms a legislature could write. "Filtering" could mean human review of reported threats or it could mean pre-publication scanning, and those have very different consequences for free expression. Until it says which, the demand is a slogan.
The case in Durban shows what the law already gives complainants: a police report, an identifiable suspect and a prosecutor. Proclaiming the 2023 Act would add to that toolkit. It should be done on its merits, without pretending it solves the platform problem.