South Africa's Electoral Commission (IEC) has drafted a new code of conduct to penalize political parties and candidates for publishing misleading election content [1].
These measures aim to protect the integrity of the upcoming 2026 Local Government Elections by curbing the spread of digital manipulation [1]. As artificial intelligence becomes more accessible, the risk of coordinated disinformation campaigns threatening voter behavior has increased [2].
The draft Electoral Code of Conduct on Countering Disinformation specifically addresses the use of AI, deepfakes, and other digital tools [1]. Under these proposed rules, the IEC would have the authority to penalize those who use these technologies to distribute false information to the public [2]. The guidelines were previously gazetted in 2024 to prepare for the current electoral cycle [2].
Officials said the rules are necessary to ensure that voters make decisions based on accurate information. The focus remains on preventing the manipulation of the electorate through sophisticated digital content that can mimic real people or events [1].
The IEC vice-chairperson said the need for these protections in the Government Gazette [1]. By establishing clear penalties for disinformation, the commission intends to hold political actors accountable for the content they produce and distribute online [2].
The move comes as governments worldwide struggle to regulate generative AI in political contexts. The IEC is positioning these rules as a shield against the potential for AI-driven instability during the voting process [1].
“The draft Electoral Code of Conduct on Countering Disinformation would penalise political parties and candidates for publishing misleading election content.”
South Africa is attempting to preemptively mitigate the 'liar's dividend'—where the existence of deepfakes allows politicians to deny real events or fabricate fake ones. By codifying these penalties before the 2026 local elections, the IEC is moving from a reactive to a proactive stance on digital governance, though the effectiveness will depend on the commission's ability to detect AI content in real-time.
