Deepfakes, investor vulnerability and trust erosion: evidence from the Czech Republic
- 3 November 2025
- journal article
- Published by Emerald Publishing in The Journal of Risk Finance
- Vol. 27 (1) , 68-84
Abstract
Purpose: The study deals with the dynamic phenomenon of deepfakes, which as demonstrated by the instance of the Czech Republic, significantly interferes with the economic ties (for instance, fraudulent investment offers when abusing the faces of business or politics). The main objective of the study was to examine the relationships between the attitudes of the population towards individual types of deepfakes and the sociodemographic characteristics of the respondents, as well as between the fields of their media consumption and personal experiences with deepfakes in the conditions of the Czech Republic and thus, to identify the risk potential of deepfakes at both the individual and societal levels. Design/methodology/approach: Data collection was carried out through the CASI approach via Ipsos online panel Populace.cz in April 2025 on a sample of 1,262 respondents. The analytical processes were carried out using logistic binomial and ordinal regression analysis. Findings: The most significant finding of the primary research was that 29% of respondents had their own experience with deepfake content, while the perception of deepfake is generally mostly negative and unaccepted. Respondents often experienced deepfakes associated with a favourable financial evaluation (59.6%). Regarding the aspects of media consumption, the preference for alternative media seems to be the most significant predictor of more positive perception of deepfake, while information fatigue and emphasis on free content were associated with a more sceptical attitude. Practical implications: These outcomes suggest that attitudes towards deepfakes are not only a reflection of the technology itself, but also a consequence of the media that people watch and how they access information. In practice, this means that when building media literacy and critical thinking, it is important to take into consideration the diversity of media preferences as these can shape not only awareness, but also trust and evaluation of new technologies such as deepfakes. Social implications: Age variability in the perception of deepfakes confirms that the different population groups may have different levels of vulnerability to the multiaspect negative effects of deepfakes (financial, emotional, psychological and reputational damage). Originality/value: The perception of deepfakes technologies is not uniform in the population, but is formed according to the personality and socio-demographic characteristics. Elderly population perceives deepfakes differently than younger people, confirming the variability in their ability to identify deepfakes. The study outcomes are beneficial for policymakers and regulatory frameworks, essential for the creation of prevention programmes and protective tools in order to minimise financial and other risks and harms related to deepfakes.Keywords
Funding Information
- Technology Agency of the Czech Republic (SIGMA Programme)
- Technology Agency of the Czech Republic (project TQ01000100)
This publication has 28 references indexed in Scilit:
- Unmasking deepfakes: a multidisciplinary examination of social impacts and regulatory responsesHuman-Intelligent Systems Integration, 2025
- Does trust in government moderate the perception towards deepfakes? Comparative perspectives from Asia on the risks of AI and misinformation for democracyGovernment Information Quarterly, 2024
- Extending the theory of information poverty to deepfake technologyInternational Journal of Information Management Data Insights, 2024
- Opposing agents evolve the research: a decade of digital forensicsMultimedia Tools and Applications, 2024
- Deepfakes and scientific knowledge disseminationScientific Reports, 2023
- Deepfakes: regulatory challenges for the synthetic societyComputer Law & Security Review, 2022
- Deepfakes and Society: What Lies Ahead?2022
- Fooled twice: People cannot detect deepfakes but think they caniScience, 2021
- Applied Logistic Regression2013
- Analysis of Ordinal Categorical Data2010