Utah Valley University’s (UVU) Center for National Security Studies’ Emerging Tech Policy Lab (EMTECH) and the Gary R. Herbert Institute for Public Policy, have published research findings that show that deepfakes are as effective as authentic content in shaping voter opinion. It further found that voters can no logner reliably identify whether they had been exposed to deepfake content, regardless prior familiarity with deepfakes, age, or political affiliation. The findings of this study have significant implications for national security, economic development, and public policy.
The study used an online survey platform to conduct testing and collect data from a politically representative sample of the United States population to examine how AI-generated content could impact voter’s opinion and election security.
Results of the study show that there is no statistical significance in the ability of a video to alter someone’s previously held opinion if it was deepfake or a legitimate video, showing that deepfake videos influence the opinions of potential voters just as effectively as real media.
Additionally, detection rates for synthetic videos were roughly equal across political affiliations — Democrats, Republicans, and Independents were only able to identify deepfake content correctly at rates between 15-19%. No age group, political affiliation, or other tested demographic showed a meaningful advantage. Deepfake detection failure is a universal vulnerability, not one concentrated in any particular group.
Familiarity with deepfakes did not improve people’s ability to recognize them. Participants who said they were “not at all familiar” with deepfakes correctly identified whether they had seen deepfake content at the same rate as those who claimed that they were "extremely familiar."
One area of particular concern is public discourse and election security. In recent years, dozens of cases of AI-enabled election disinformation have emerged around the world, affecting contests ranging from federal elections to local and municipal ballot initiatives. However, the extent of their impact remains widely debated. Individuals targeted by deepfakes often argue that the content substantially influenced election outcomes, while electoral beneficiaries of deepfake disinformation frequently minimize its significance.
“The findings of this study present a clear and urgent challenge for policymakers, election security professionals, and the public: AI-generated deepfakes are already capable of influencing political opinion as effectively as authentic media, and the people most likely to encounter them are not equipped to identify them,” said Brandon Amacher. “Equally concerning is the finding that familiarity with deepfakes provides no meaningful protection. This undermines a common assumption embedded in current media literacy policy — that educating the public about the existence and nature of deepfakes will reduce their impact.”
The study was designed to replicate the way individuals are likely to encounter deepfakes in real-world settings, particularly while casually scrolling through social media without heightened suspicion or awareness. To mirror these conditions, participants were not informed at the outset that they might view AI-altered content. Instead, they were told that the study examined how individuals form and adjust opinions after viewing short-form media content. Only after subjects were shown a media sample and measurements about the impact of the media on the viewer were taken, was the true nature of the study revealed.
All deepfake materials used in the study were made by students at the UVU Center for National Security Studies using publicly available tools. In order to mitigate the introduction of potential biases, we attempted to choose an issue that would not elicit strong emotional reactions. The text that appears in the survey (a fictional federal housing assistance initiative) was written to resemble an issue which might appear on a real ballot.
EMTECH operates at the intersection of cutting-edge technology, public policy, and national security. Serving as both a consulting firm and a think tank, EMTECH addresses some of the most urgent and complex challenges posed by emerging technologies.
EMTECH conducts a broad range of projects for partners across the intermountain region and nationwide, including collaborations with federal agencies, state and local governments, and private sector organizations. Our team combines policy expertise, technical proficiency, and strategic insight to deliver actionable solutions that strengthen infrastructure resilience and inform security policy at all levels.
To read the complete study, please visit https://www.uvu.edu/herbertinstitute/docs/brandonamacherpaper2026_web.pdf.
