Deepfakes and Disinformation: Exploring the Impact of Synthetic Political Video on Deception, Uncertainty, and Trust in News

By Cristian Vaccari, Andrew Chadwick. 2020

824 citations recorded by the source corpus.

Artificial Intelligence (AI) now enables the mass creation of what have become known as “deepfakes”: synthetic videos that closely resemble real videos. Integrating theories about the power of visual communication and the role played by uncertainty in undermining trust in public discourse, we explain the likely contribution of deepfakes to online disinformation. Administering novel experimental treatments to a large representative sample of the United Kingdom population allowed us to compare people’s evaluations of deepfakes. We find that people are more likely to feel uncertain than to be misled by deepfakes, but this resulting uncertainty, in turn, reduces trust in news on social media. We conclude that deepfakes may contribute toward generalized indeterminacy and cynicism, further intensifying recent challenges to online civic culture in democratic societies.

Subfields: Sociology and Political Science, Statistical and Nonlinear Physics, Evaluation & Benchmarks, Safety, Robustness & Interpretability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Sociology and Political Science

In Statistical and Nonlinear Physics

In Evaluation & Benchmarks

In Safety, Robustness & Interpretability

Citations in the corpus

Evaluating the Social Impact of Generative AI Systems in Systems and Society (2023, background): “What to Evaluate Surveying trust can apply to trust in AI systems [184, 107] to output factual information, trust in researchers, developers, and organizations developing and deploying AI [143], mitigation and detection measures [222], and trust in overall media and how it is distributed [251].”
Score-based Generative Modeling in Latent Space (2021, background): “Here, LSGM is examined on the image generation task which has potential benefits and risks discussed in [94, 95].”
An Overview of Catastrophic AI Risks (2023, background): “Alternatively, if AIs are broadly accessible this could lead to widespread disinformation, with people retreating to trusting only a small handful of authoritative sources [23].”
Interpretable and Trustworthy Deepfake Detection via Dynamic Prototypes (2021, background): “…– which can be utilized to alter the identity of a person in a video – have emerged as perhaps the most sinister form of misinformation, posing a signiﬁcant threat to communities around the world [63, 26, 60, 62], especially with election interference or nonconsensual fake pornography [9, 24].”
Controllable and Compositional Generation with Latent-Space Energy-Based Models (2021, background): “On the application side, it shares with other image synthesis tools similar potential benefits and risks, which have been discussed extensively in [5, 52].”
ChatGPT and the rise of large language models: the new AI-driven infodemic threat in public health (2023, background): “Political Deepfakes already contributed toward generalized indeterminacy and disinformation (67).”
EditGAN: High-Precision Semantic Image Editing (2021, background): “The recent progress of generative models and AI-driven photo editing has profound implications on image authenticity and beyond, which is an area of active debate [83].”
Score-Based Generative Modeling with Critically-Damped Langevin Diffusion (2021, background): “Generative modeling of images has promising applications, for example for digital content creation and artistic expression (Bailey, 2020), but can also be used for nefarious purposes (Vaccari & Chadwick, 2020; Mirsky & Lee, 2021; Nguyen et al., 2021).”
LION: Latent Point Diffusion Models for 3D Shape Generation (2022, background): “Although we do not see any immediate negative use-cases for LION, it is important that practitioners apply an abundance of caution to mitigate impacts given generative modeling more generally can also be used for malicious purposes, discussed for instance in Vaccari and Chadwick [122], Nguyen et al.”
HyperStyle: StyleGAN Inversion with HyperNetworks for Real Image Editing (2022, background): “While our tool aims to empower content creators, it can also be used to generate more convincing deep-fakes [25] and aid in the spread of disinformation [27].”