Repeated exposure to deepfakes makes you more likely to believe their claims, finds an international study by NTU Singapore.
According to a false video generated using artificial intelligence that went viral on social media, American media personality Kim Kardashian manipulates people online for money, and you are more likely to believe this if you come across the same video again online, according to a multi-country study by Nanyang Technological University, Singapore (NTU Singapore).
Through an international study of 8,070 study participants, the NTU Singapore team found that those who had seen this viral deepfake of Kim Kardashian on social media prior to the study were more likely to believe the claim made in the video when exposed to it again.
The same effect was replicated when the study participants from Singapore, China, Indonesia, Malaysia, the Philippines, Thailand, the United States, and Vietnam were shown viral deepfakes of three other well-known figures as part of the study: Mark Zuckerberg, Vladimir Putin, and Tom Cruise.
The scientists suggest that this is due to the illusory truth effect, a psychological phenomenon in which people are more likely to believe information is true simply because they have encountered it multiple times, regardless of its accuracy. Repetition increases familiarity with the information, which in turn makes it easier for the brain to process.
This effect is amplified in study participants who rely primarily on social media platforms for news rather than direct news sources (e.g. websites, TV, newspapers) and thus are more likely to encounter deepfakes, increasing their risk of accepting the false claims as true. Contrary to existing research, the scientists found that having a higher cognitive ability did not defend these participants against the amplified illusory truth effect.
“Our study shows that across participants from different socio-political contexts, repeated exposure to the same deepfake makes them more likely to believe it, suggesting that the illusory truth effect persists regardless of their cultural and political background,” said the NTU’s Assistant Professor Saifuddin Ahmed. “As deepfakes become increasingly common, there is a pressing need for governments, tech companies, and media outlets to collaborate on solutions that mitigate their impact.”
“Given that mere exposure to deepfakes could reinforce false beliefs, policymakers could consider psychological mechanisms (such as the illusory truth effect) when developing educational campaigns focused on debunking deepfakes,” he added. “Policymakers could also make use of the same psychological mechanism to disseminate important and truthful information, such as during a health crisis, to subconsciously encourage certain behaviours in the public.”
The NTU scientists also looked at the national differences in the perceived believability of deepfakes.
Respondents in Singapore were the least likely to be deceived by deepfakes among the eight countries, followed by Vietnam and the Philippines.
The scientists said this could be due to higher levels of digital literacy in Singapore and a sustained effort from the government to raise awareness against misinformation through proactive campaigns and legislation.
However, the scientists noted that this observation does not necessarily mean Singaporeans are the most resilient to deepfakes.
“In fact, a large majority of the study respondents from Singapore who did not believe the false claims in the deepfakes shown were also found to be uncertain about the accuracy of the claims. They felt ‘it may be true or not’.” Saifuddin said. “While this is better than accepting the misinformation, it may reflect a broader sense of wariness towards unfamiliar information online. This may leave individuals more vulnerable to future exposure, as repeated misinformation could exploit their indecision”.
“Uncertainty also limits their ability to counter misinformation in social settings, allowing false narratives to spread. In contrast, certainty-based rejection builds cognitive immunity, making individuals more resistant to future persuasion and better equipped to challenge misinformation actively.”
The study was published in the Journal of Broadcasting & Electronic Media.
