![[CAPTION] South Africa lacks the election transparency tools available to voters in stricter jurisdictions. (Image created via ChatGPT)](https://lh3.googleusercontent.com/dnC2uJ1Cff00om5UPHL1GgH1OXS4SmApaAkpVuAriVSCBY1g_CdYGrVuRO7bRRA6nKbiF5SUJNgkSgB9_BA_ZotYJRTnAUaQpTlP)
South African voters are set to enter the 2026 local government elections with fewer safeguards against online manipulation, voter suppression and discriminatory political targeting than their counterparts in many democracies across the globe.
This is according to a new report by the Campaign on Digital Ethics (CODE), a Johannesburg-based non-profit civil society organisation that advocates for digital rights, tech accountability, ethical technology governance and the fight against online disinformation.
The report, titled: “The algorithmic ballot: Political advertising and voter protection in the age of Big Tech”, argues that the next municipal elections will be contested not only through traditional campaigning, but increasingly through the algorithms, recommendation systems and social media platforms operated by global technology companies.
The next local government elections will be held on 4 November.
The findings of the CODE report were released during a webinar last week. Kristen Abrahams, legal and policy lead at CODE, noted the study highlights a growing disconnect between the online protections offered to voters in highly-regulated jurisdictions and those available to South Africans ahead of the 2026 local government elections.
According to the report, political influence campaigns are increasingly shaped by the systems of social media platforms such as Meta, Google, X and TikTok, which determine what content users see, which narratives gain traction and which communities are targeted with specific political messages.
“South Africa’s 2026 local government elections will be fought in two arenas: on the streets, and inside the recommendation systems of a handful of global technology companies,” Abrahams warned, unpacking the findings.
“The same vote that is canvassed through door-to-door campaigns and community meetings will also be contested through search results, social feeds, short-form videos and influencer partnerships.”
The report uses a qualitative policy analysis methodology based primarily on a review of publicly available data from the major platforms – Google, Meta, TikTok and X – focusing on their political advertising rules, transparency tools, verification systems and content moderation policies.
Regulatory inequality
A key finding of the report is what CODE describes as “regulatory inequality”. While users in regions such as the EU, UK and US benefit from stronger transparency rules, election safeguards and platform accountability measures, South African users receive a lower standard of protection.
The report notes that transparency tools made available to researchers, regulators and civil society groups in Europe are often unavailable or significantly less effective in SA.
“This analysis by CODE shows that South African voters are entering this election with weaker digital protections than voters in more tightly-regulated jurisdictions, such as the United Kingdom, and European Union,” notes Abrahams.
The report further reveals that South African researchers and platform users do not enjoy the same level of access to advertising data and transparency mechanisms that have been mandated elsewhere.
“Transparency tools that regulators have forced into existence for EU users and researchers, on platforms like TikTok, are not available at the same level of depth or reliability to South African civil society, journalists and academia. This creates a situation where harmful campaigns can operate with reduced scrutiny,” she continued.
This means online abuse and manipulation, for instance, is more likely to stay hidden. In places with stronger transparency obligations, problematic campaigns can be spotted, documented and used as evidence to pressure platforms or regulators.
Fragmented safeguards
The report unpacks how each social media platform handles political advertising and user protection with different approaches – an element that leads to varying levels of exposure for South African users.
“Meta allows political adverts but requires advertiser verification, ‘paid for by’ disclosures and logs them in a public ad library; however, its protections are more fully enforced in tightly-regulated regions like the EU than in SA.
X (formerly Twitter) allows political advertising in approved markets under specific campaigning rules – requiring advertisers to be based in the country they target – but public trust in its consistent enforcement remains weaker than its written policy.
“Ultimately, because platforms enforce their policies much more stringently where regulatory pressure is tougher, South African social media users remain highly-vulnerable to risky tactics and targeted misinformation.
“TikTok implements the strictest written ban by completely prohibiting paid political advertising and compensated branded political content for parties and politicians. However, concerns persist regarding its enforcement quality and a potential shift toward undisclosed influencer content often used by the platform to spread political influence.”
Micro-targeting risks
Among the study’s strongest warnings is the growing use of micro-targeting in political advertising.
Micro-targeting enables political campaigns to use personal information and behavioural data to deliver highly-tailored messages to specific groups of voters, via SMSes or e-mail.
The report explains that a political party could simultaneously present completely different messages to different communities without any public visibility into those strategies.
“Imagine Party X showing one message to young men in Soweto, a completely different message to Indian voters in Lenasia, and nothing at all to voters whom it has already written off. When harmful or misleading messages are tightly micro-targeted to narrow, specific groups, they can cause significant damage before anyone outside that specific audience segment even knows they exist,” the report cautions.
CODE argues that such practices are particularly concerning in SA because of the country’s history of apartheid spatial planning.
Furthermore, micro-targeting could make it hard for journalists, researchers and watchdogs to identify or fact-check the misinformation, as they only see isolated fragments of the broader ad universe.
According to the report, most platforms are implementing rules that require political advertisers to use self-reporting checkboxes, or mandatory disclosures to flag AI-generated and synthetic content.
However, these efforts rely heavily on creator honesty and voluntary frameworks, rather than robust, automated detection systems, leaving enforcement uneven and heavily dependent on localised regulatory pressure.
Transparency gaps remain
The report draws parallels with the Cambridge Analytica scandal, which became one of the most significant examples of data-driven political manipulation.
CODE recounts how the data analytics company harvested personal information from millions of Facebook users and used that information to build detailed voter profiles.
Quoting whistle-blower Christopher Wylie, the report notes the resulting system was described as a “psychological warfare tool”.
According to the report, the scandal demonstrates how digital platforms can be used to identify emotional vulnerabilities and exploit them for political purposes.
It also highlights shortcomings in political advertising transparency across major platforms.
While Meta and Google maintain public advertising archives, CODE argues that South African users still receive less visibility than users in more heavily regulated regions.
“You see the ads, not the system. In SA, you can sometimes look up political ads (especially on Meta or Google), but you do not get the full, structured view which researchers and regulators have pushed for in the EU.
“That makes it harder to understand how big a campaign really is, who it is targeting and how much money is being spent behind the scenes.”
The emergence of generative AI also presents another challenge ahead of the elections.
Although major platforms have introduced policies requiring disclosure of AI-generated content, CODE argues that enforcement remains inconsistent and heavily dependent on self-reporting.
“The absence of an AI label should not be interpreted as proof that content is authentic. You cannot assume ‘no label’ means ‘no AI’. A political image or video without an AI label may still be synthetic or heavily manipulated. The absence of a disclosure often reflects a failure of enforcement or bad faith non-compliance, not authenticity.”
CODE believes the threat extends beyond individual cases of misinformation.
“The democratic risk is structural, not individual. This is not just about whether one user is fooled by one video. It is about entire communities seeing fabricated evidence that a candidate said or did something they never did – and potentially losing faith in the possibility of knowing what is real at all,” the report states.
CODE says it is using these findings to actively engage with the IEC. This comes at a crucial time, as the IEC has announced plans to issue a draft Code of Conduct on Misinformation for public comment to protect information integrity.
Share
