Most people worry about deepfakes – and overestimate their ability to spot them

The rise of generative artificial intelligence (gen AI) is colliding with elections all over the world, and consumers are worried.

Singapore, Mexico, the US, and the UK are all holding general elections this year — and voters are concerned they will not be able to identify deepfakes. According to a report released by digital identity management vendor Jumio, most consumers are anxious about the potential for AI and deepfakes to influence upcoming elections. 

Conducted in late March, the study surveyed 8,077 total respondents, roughly 2,000 from each of the four countries. The UK held its general elections early this month, while Mexico had its own last month. The US will hold its presidential elections in November, and Singapore is anticipated to hold its general elections later this year or before November 2025 at the latest. 

Also: All eyes on cyberdefense as elections enter the generative AI era

The study found that 72% of respondents worry daily that they may be fooled by a deepfake, with those in Mexico and Singapore feeling most concerned at 89% and 88%, respectively. In comparison, 57% in the UK and 55% in the US reported having the same worry. 

About 60% of respondents believe they have encountered a deepfake over the past year, while 22% are uncertain if they have. Just 15% say they have yet to come across a deepfake.

Despite being worried they may be fooled by a deepfake, 60% are confident they can detect one. Jumio said this figure is 8% higher than last year. Respondents in Singapore are the most confident at 77%, compared to 58% in the US, 54% in Mexico, and 53% in the UK.

Also: 80% of people think deepfakes will impact elections. How to prepare

Specifically, 46% of global respondents believe they can identify a deepfake of a politician. Singaporeans are the most certain of each country at 60%, followed by Mexico at 51%, the US at 37%, and the UK at 33%. 

Globally, 71% say they are more skeptical of online political content compared to the last election — although 43% of respondents said they trust the political news they see online, despite the possibility of encountering deepfakes.

When asked if they agreed that their government’s laws around AI don’t do enough, 60% across the board said yes. The majority of Singaporeans agreed with the highest percentage of all four countries at 67%, compared to the lowest at 53% in the UK. 

Also: Can governments turn AI safety talk into action?

Singapore respondents, though, express the most faith in their government’s ability to regulate AI, at 69%. In comparison, just 26% in the UK are confident in their government’s ability to regulate AI, alongside 31% in the US and 44% in Mexico. 

“With half of the global population participating in elections this year, the potential influence and impact of generative AI and deepfakes demand our immediate attention,” Jumio CEO Robert Prigge said. “Public faith in online information is crumbling, demanding a transparent discourse to confront this challenge and empower citizens with the tools to discern and report deepfakes.”

Also: More political deepfakes exist than you think, according to this AI expert

Prigge added that online platforms have “a critical duty” to implement detection measures, such as multimodal, biometric-based verification systems, to combat deepfakes that could potentially influence elections.

Jumio noted that social media platforms in Indonesia carried fake content prior to the country’s February elections, including a video depicting deceased former president Suharto endorsing the incumbent. The cybersecurity vendor added that Singapore is mulling over a temporary ban on political deepfakes in the leadup to its elections, while South Korea adopted a 90-day ban on the use of deepfakes in political campaign content.

READ MORE HERE