Electoral Commission Warns of Unfinished AI Safeguards Before November 2026 Vote

SOUTH AFRICA FACES NOVEMBER ELECTIONS WITH AI SAFEGUARDS STILL INCOMPLETE

The Electoral Commission of South Africa confronts a regulatory vacuum as the country prepares for local elections in November 2026. The IEC itself has flagged the central governance challenge: artificial intelligence, particularly generative AI deployed in hyperlocal contexts, poses a novel threat to electoral integrity that existing law was not designed to address.

IEC Chair Mosotho Moepya identified the problem plainly. “The primary challenge of 2026 is the rise of generative AI and its application in ‘hyperlocal’ contexts,” he said, noting a shift from broad national falsehoods to ward-specific deceptions. The IEC expects a “flurry of deepfakes” in the coming elections, citing warnings from the South African National Editors’ Forum.

The governance problem is structural. South Africa does not have a dedicated AI statute. It lacks an AI-specific framework for elections. Instead, regulation is scattered across multiple instruments that predate generative AI and were not built to address synthetic media or algorithmic manipulation. The Municipal Electoral Act criminalizes false election-related information intended to disrupt polls, but contains no provisions for deepfakes. The Cybercrimes Act offers a criminal backstop for malicious digital disinformation. The Protection of Personal Information Act governs the data collection underlying AI-driven micro-targeting. The National AI Policy Framework signals future direction but remains a policy proposal without binding legal force, and will have no effect during the 2026 elections.

The IEC’s draft regulations, published in June 2026, represent the first election-specific AI response. They propose that political parties label AI-generated campaign material, verify and correct false claims, report online disinformation and assume responsibility for content on their own platforms. These are regulatory measures relying on codes of conduct and self-reporting, not legislation.

The fragmentation matters most in a specific context. South Africa’s political landscape has become the most divided since 1994. Dominant party politics has declined. Numerous smaller parties have emerged, independent candidates wield growing influence, and coalition politics intensifies. Electoral margins in many wards are narrow. In this environment, a hyperlocal deepfake circulated through a single ward WhatsApp group two days before polling could shift a few hundred votes and determine which party holds a deciding seat on a hung council.

The scale of the vulnerability is large. The IEC expects more than 100,000 candidates to contest over 4,400 wards, exceeding the approximately 95,000 who stood in 2021. Coalition governments at the municipal level already tend toward volatility and personality-driven decision-making. Layering a low-cost, high-volume disinformation tool onto that instability raises the stakes of ward-level manipulation far beyond what dominant-party politics would have entailed.

The IEC’s own framing acknowledges the governance limits. “The integrity of our 2026 local government elections does not rest on the IEC alone,” Moepya said. “It rests on the fact-checker in Johannesburg, the legislator in Cape Town, the tech engineer in Silicon Valley and the EU diplomat in Pretoria.” This is an admission that transparency, media literacy and platform cooperation are bearing as much responsibility as statute.

By contrast, comparable jurisdictions offer limited reassurance. The European Union’s AI Act imposes binding, risk-tiered obligations on high-impact systems, yet election-specific enforcement has proved slower and messier than the statute’s ambitions suggested. Brazil’s electoral court took a more interventionist route, imposing binding AI-specific campaign rules before its own polls, judging that general data or cybercrime law would move too slowly within an election cycle. The IEC’s draft code gestures toward Brazil’s model but lacks legislative force.

The broader lesson is not that South Africa lags behind a settled global standard. No jurisdiction has fully solved the problem of matching an election’s compressed timeline to the naturally slower pace of statute and enforcement. The interim solution appears to rest on civil society coalitions, rapid-response fact-checking networks and platform-level content moderation agreements functioning as stopgap measures while formal legislation develops. Voluntary cooperation, though, is inherently fragile, dependent on goodwill and resourcing that can evaporate under political pressure.

Should the 2026 elections proceed without major incident, that outcome is likely to reflect the resilience of these informal networks rather than any structural fix to the underlying regulatory gap. Until law catches up, the burden of protecting electoral integrity will continue to fall disproportionately on nonstate actors and voluntary governance mechanisms. This is a constitutional reality that is necessary and, ultimately, unsatisfactory. The more pointed question, one the IEC has not yet answered, is what happens to that informal architecture if it is tested hard and fails.