From South Africa to Mexico: How AI is shaping elections worldwide

Disinformation and its potential to destabilise

In South Africa, AI's role in the electoral process remains limited, but the risks are still there. Lindiwe Mazibuko of Futurelect, from South Africa, noted that while AI tools like generative models and deepfakes had little impact on the recent elections, the country is far from immune to the dangers of disinformation. "The biggest challenge in terms of disinformation that we faced was around fears about the legitimacy and the freedom and fairness of the election," she said. While South Africa has not seen the same AI-driven electoral interference as other countries, the use of social media to erode trust in the democratic process remains a threat, partcularly if generative AI models improve in South Africa’s dominant languages.

Manipulation of young voters in Indonesia

Indonesia presented a different picture. Ika Ningtyas, Fact-Checking Coordinator at Tempo, highlighted how AI was used extensively in the lead-up to the country’s elections. "Predictive and generative AI were used widely by both presidential and legislative candidates," she explained.

These tools helped create targeted campaigns, especially aimed at engaging younger voters. "About 56% of Indonesian voters are under 40 years old," Ika said, pointing out that AI was used not to create deceptive content but rather to generate fun, entertaining content on an unprecedented scale to capture this demographic’s attention and ‘reputation wash’ candidates. But behind these flashy campaigns, there was a darker side. "The new strategy is to avoid supporting human rights," she warned, raising concerns about AI’s ability to manipulate and distract from critical political issues, including Indonesia’s long-standing human rights challenges.

Need for new electoral laws in Mexico

Cristina Martínez Pinto, from Mexico’s PIT Policy Lab, offered further insight into how AI is impacting democracy. In the Mexican elections, AI was used for micro-targeting voters, spreading disinformation, and even creating deepfake videos of political candidates.

Pinto shared an alarming example: "A fake video featuring the now-elected president promoting a Ponzi scheme was used to undermine her credibility," she recounted. While the front-runner still won, the incident underscores how easily AI can be used to manipulate public opinion and distort the democratic process. Pinto added that this use of AI has prompted calls for new electoral laws to regulate the use of AI-generated content in political campaigns.

India’s battle with misinformation

In India, the scale of AI’s impact is still uncertain, but Patil noted that the country has already been grappling with misinformation for years. "There’s so much misinformation that it’s hard to know what the counterfactual is – did it work or not?" he said.

He argued that while AI-generated misinformation has not yet had a significant impact, India’s deeply entrenched misinformation networks, often amplified by social media, remain a concern. AI could make it easier to produce and distribute false information at a much larger scale, Patil explained, adding, "We need to develop immunity to misinformation."