Scrolling to the Polls: Inside South Africa’s New Election Reality
A new survey by the Campaign on Digital Ethics (CODE) finds widespread concern about false information ahead of the 2026 local government elections.
As South Africa prepares for the local government elections on 4 November, voters are entering an electoral environment in which social media has become almost impossible to separate from political life.
In a survey of 771 people conducted by CODE in July 2026, 83% of respondents said false information posed a serious threat to South Africa’s democracy, while 60% were concerned that AI-generated content could mislead voters during the 2026 elections.
The findings form part of CODE’s new report, The Algorithmic Ballot: Volume II – Democracy in the Feed, which examines how people encounter and assess election information and considers whether South Africa’s newly introduced Electoral Code of Conduct on Countering Disinformation is equipped for the digital realities of modern campaigning.
Download the report
While the survey is not nationally representative, the results can still be read as an account of the experiences of this particular group.
Sixty percent of respondents said they had encountered rumours intended to provoke fear or anger between different social groups. Forty-four percent had encountered a false news website designed to appear legitimate, 39% said they had seen an AI-generated representation of a politician, and 26% had encountered an account impersonating a political party, candidate or the Electoral Commission of South Africa (IEC).
Nearly a quarter had encountered false claims about voting dates, locations or procedures.
The report’s most significant finding concerns where people now encounter political information. Some 91% of respondents obtained election-related information from at least one social or messaging platform. Broadcast media remained important, with 73% selecting television or radio, but the findings suggest that digital platforms have become central infrastructure through which political information travels.
Widespread use, however, does not mean widespread trust.
Only 11% of respondents expressed high trust in social media platforms as sources of information. Traditional radio and television performed considerably better, with 45% expressing high trust, followed by the IEC at 32% and fact-checking organisations at 30%.
That creates an unusual democratic contradiction where voters increasingly encounter politics through information systems they themselves regard as unreliable.
The problem is compounded by the way information moves between online and offline relationships. A claim received from a relative, neighbour, church group or community network may arrive through WhatsApp as a forwarded video or voice note, allowing the trust attached to a personal relationship to transfer to information whose original source is unknown.
Voters know more than we may assume
The survey does not, however, depict South Africans as passive victims of disinformation. Respondents demonstrated a relatively strong foundation of digital literacy. Eighty-three percent correctly identified missing information such as an author, publication date and contact details as warning signs that a news website might be unreliable, while 80% understood what a “Paid Partnership” label meant.
Seventy-eight percent knew that an unverified WhatsApp message claiming that their voting station had changed should first be checked against legitimate IEC sources, and 75% correctly identified what should be verified before sharing a viral political video. Half of respondents answered all four of CODE’s practical literacy questions correctly.
The difficulty emerged as verification became more demanding.
People were more likely to perform quick checks, such as looking at the date of a post, than undertake tasks requiring additional time, data or technical knowledge, such as tracing the original source of material, comparing reports across sources or conducting reverse-image searches.
That gap matters because sophisticated disinformation is often designed to provoke an emotional reaction before verification takes place.
Young voters face a particular challenge
The findings also point to an important challenge among younger voters. Among respondents aged 18 to 24, 34% said they had previously shared something that was later found to be false or misleading, compared with 29% among those aged 25 to 34 and 20% among respondents aged 35 to 54.
At the same time, registration and intended electoral participation were lower among younger respondents. CODE cautions that this does not mean young people are inherently more susceptible to disinformation. Instead, the findings suggest that their political participation is taking place within a highly complex information environment while their relationship with formal democratic institutions remains comparatively weak.
Political influence is becoming harder to identify
Another concern is the increasingly blurred boundary between political opinion and paid political promotion. Sixty percent of respondents said they had seen an influencer or content creator promote a political party or candidate, while 63% had encountered political promotion on social media that they suspected had not been clearly disclosed as advertising.
Seventy-one percent believed influencers should be legally required to disclose paid political promotion.
For voters, however, establishing whether an influencer is independently expressing a political view, being paid by a party or working through an intermediary can be extremely difficult.
A new legal test for South Africa
The upcoming local government election will also provide the first major test of South Africa’s Electoral Code of Conduct on Countering Disinformation, which came into force on 4 September.
The Code creates binding obligations for political parties and candidates and covers conduct carried out by people acting on their instruction, authority or for reward. Political actors are required to take reasonable steps to check information before publishing it and are prohibited from deliberately or recklessly spreading false information.
It also addresses deceptive digital practices including fake accounts, automated bots, misleading synthetic content and the misuse of sensitive personal information.
Political advertising must identify the relevant party or candidate, while AI-generated material must be labelled as synthetic. Where false or misleading information has been disseminated, parties and candidates must publicly retract and correct it as soon as reasonably possible and within 36 hours of becoming aware of it.
Real411, a platform developed by Moxii Africa (formerly Media Monitoring Africa) has been designated as the official online mechanism for reporting suspected election-related disinformation.
Importantly, the Code also protects lawful criticism, opinion, journalism, satire and legitimate political debate, which is an attempt to prevent rules designed to tackle disinformation from becoming a mechanism for suppressing disagreement.
CODE argues that the new framework nevertheless contains significant gaps.
Its binding obligations primarily apply to political parties, candidates and people demonstrably acting on their behalf or for reward. Independent influencers, podcasters, commentators and other organisations may therefore fall outside its direct reach unless a relationship with a party or candidate can be established.
More significantly, the Code places no direct obligations on the technology companies whose systems distribute political information.
Platforms are not required under the Code to maintain public repositories of political advertising, disclose how advertisements were targeted, reveal which accounts paid to reach particular audiences or provide researchers and regulators with information about the spread of election-related content.
Nor does the Code require a correction to reach the same audience that saw the original false claim.
This creates what CODE describes as a fundamental mismatch: 91% of its respondents encounter election information through social or messaging platforms, yet the companies controlling how that information is amplified acquire no direct obligations under the new instrument.
Voters want the system to do more
The survey suggests respondents themselves do not believe digital literacy alone will solve the problem. Half wanted a simpler way to report or fact-check suspicious content. Forty-four percent wanted clearer labels on political advertising and AI-generated material, while 35% wanted short practical guides or videos and 32% wanted workshops or community training.
Eighteen percent identified lower data costs when verifying information as something that would help them. The report argues that this raises an overlooked question of inequality: misleading material may be cheap or free to receive and forward, while opening news reports, watching explanatory videos, performing reverse-image searches or accessing reporting portals can consume paid data.
CODE is consequently calling for a response that combines digital literacy with stronger accountability across the information chain, including greater transparency around paid political influence, direct obligations on platforms and publicly visible monitoring of how the new Disinformation Code is enforced.
The organisation argues that corrections should also travel through the same channels, with comparable visibility, as the false information they are intended to remedy.
The November election will provide the first opportunity to see whether South Africa's new framework can keep pace with the information environment voters actually inhabit.
The deeper question raised by CODE's research is no longer simply whether voters can identify a false post. It is whether the wider digital system in which political choices are being formed gives them a fair chance to do so.
Comments