Shubham Bhatia

https://doi.org/10.64823/ijcml.2601001

The rapid diffusion of generative artificial intelligence (AI) has made synthetic political media deepfakes and cheapfakes an increasingly salient concern for democratic communication. This study examines public awareness, perception, skepticism, and self-reported behavioral response to AI-generated political content among a diverse sample of 404 respondents. Using a structured survey instrument covering demographic profile, awareness, individual perception, decision-making impact, media skepticism, and democratic participation, the study finds near-universal awareness of AI (99.5%) but comparatively lower recognition of specific terminology such as ‘deepfake’ and ‘cheapfake’ (85.6%). One-sample chi-square goodness-of-fit tests confirm that response distributions on every attitude item depart significantly from chance (p < .001 in all but two cases), indicating that respondents hold clear, non-random positions rather than indifferent ones. Two-proportion z-tests show that respondents overwhelmingly agree that AI-generated media undermines the trustworthiness of political information (58.4% vs 31.2%, z = −5.27, p < .001), and that exposure has made them more cautious in decision-making (70.8% agreement, z = 14.40, p < .001). A recurring third-person pattern emerges: respondents report personal resilience to influence (61.9% say their own opinions are unaffected) while simultaneously reporting that AI-generated media erodes trust and discourages public dialogue at a societal level (69.8–73.0% agreement). The findings suggest a public that is highly aware but deeply skeptical, one that is adapting through heightened vigilance rather than blind acceptance, with clear implications for platform labeling policy, media literacy interventions, and electoral regulation.