Student or educator? Sign up with your school email and get 6 months of Pro free.

The DOI matches its Crossref record; 62 of the 70 references it cites were found in a registry or an index, and 8 are not checked yet; 7 of the 54 sources its claims cite were checked for support.

The detection of political deepfakes

Open access
Appel, Markus, Prietzel, Fabian·2022·Journal of Computer-Mediated Communication

Abstract

Deepfake technology, allowing manipulations of audiovisual content by means of artificial intelligence, is on the rise. This has sparked concerns about a weaponization of manipulated videos for malicious ends. A theory on deepfake detection is presented and three preregistered studies examined the detection of deepfakes in the political realm (featuring UK’s Prime Minister Boris Johnson, Studies 1–3, or former U.S. President Barack Obama, Study 2). Based on two system models of information processing as well as recent theory and research on fake news, individual differences in analytic thinking and political interest were examined as predictors of correctly detecting deepfakes. Analytic thinking (Studies 1 and 2) and political interest (Study 1) were positively associated with identifying deepfakes and negatively associated with the perceived accuracy of a fake news piece about a leaked video (whether or not the deepfake video itself was presented, Study 3). Implications for research and practice are discussed.

Verified 2026-10-06 · 8 references not checked yet

Overview

CoChat Summary

TL;DR

Three preregistered German online studies found that analytic thinking (Studies 1–2) and political interest (Study 1) predicted better detection of political deepfakes, while a deepfake embedded in a fake news article did not increase perceived accuracy or alter leadership ratings (Study 3).

Key findings

  • In Study 1, 82 of 199 participants (41.2%) stated or suspected that the Boris Johnson deepfake was fake; analytic thinking (OR = 1.23, p = .013) and political interest (OR = 1.49, p = .025) were positively associated with detection.
  • In Study 2, the Johnson deepfake was rated more manipulated (M = 5.34, SD = 1.70) than the Obama deepfake (M = 4.91, SD = 1.99), both above the scale midpoint; analytic thinking predicted Johnson detection (B = .13, p = .026) and content-based detection (B = .21, p = .001), but not Obama detection, and political interest was not a significant predictor.
  • Study 2 found an indirect effect of analytic thinking on deepfake detection through implausible content identification (effect = .077, 95% CI [.015, .143]), but not through technological glitches (effect = .005, 95% CI [−.019, .028]); technological-glitch recognition was unrelated to analytic thinking, r(231) = .027, p = .680.
  • In Study 3, adding the deepfake did not affect perceived accuracy of the fake news article (M = 3.13 vs. 2.98, p = .290) or perceived leadership of Boris Johnson (M = 3.68 vs. 3.60, p = .537); analytic thinking (B = −.14, p = .045) and political interest (B = −.40, p < .001) predicted lower accuracy ratings irrespective of deepfake exposure.

Methods

  • Three preregistered online studies with German samples recruited via Clickworker.de: final N = 199 (Study 1), 233 (Study 2), and 445 (Study 3); stimuli were a Boris Johnson deepfake (Studies 1–3) and an Obama deepfake (Study 2).
  • Study 1 used an open-ended thought-listing task; Study 2 used closed-ended deepfake detection and indicator ratings; Study 3 embedded the Johnson deepfake in a fake news article versus a news-only control.
  • Predictors were a seven-item Cognitive Reflection Test for analytic thinking and a single political-interest item; analyses included logistic regression, linear regression, mediation, and ANOVA.

Limitations

  • Stated: the studies relied on stimuli available in 2020, largely one Boris Johnson deepfake; the Obama deepfake may be perceived as satirical or entertaining; self-report detection measures may inflate detection because mentioning manipulation cues respondents; future research should use unobtrusive measures.
  • Stated: context information was not an empirical focus, and the authors note that warning citizens about deepfakes could unintentionally increase distrust of real content.
  • Evident: the German samples and specific political figures limit generalizability; Study 2 gave mixed support for political interest, and Study 3’s null deepfake effects contrast with prior work by Hwang et al. (2021).

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1093/jcmc/zmac008

Cites

62 of 70 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 Graphical representation of the deepfake detection model.
Figure 1 Graphical representation of the deepfake detection model.CC-BY · From the PDF
Image 2
CC-BY · From the PDF
Figure 3 Plot of conditional deepfake detection probabilities (Study 1) .
Figure 3 Plot of conditional deepfake detection probabilities (Study 1) .CC-BY · From the PDF
Image 4
CC-BY · From the PDF

Verification

Claims the paper makes

6 Valid · 1 Not valid · 32 Not checked · 81 Own results and interpretations

47 more claims

Create a free account to see all 120 claims and to verify the 44 references still open.

See all claims — free account
  • Probably not a substantial claim. 38
  • It may state two separate claims as one. 37
  • The quote only partly states this claim. 4

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

5 more claims about this paper

Create a free account to see every claim other papers make about this one.

See all claims — free account

Citations

Cites

Show all 64 references

Research works printed but not linked: 5

  1. Cole, 2018, We are truly fucked: Everyone is making AI-generated fake porn now
  2. Goodfellow, 2014, Advances in Neural Information Processing Systems, 2672
  3. Kahneman, 2011, Thinking, fast and slow
  4. Mervosh, 2019, Distorted videos of Nancy Pelosi spread on Facebook and Twitter, helped by Trump
  5. Spivak, 2019, “Deepfakes”: The newest way to commit one of the oldest crimes, The Georgetown Law Technology Review, 3, 339

Cited by

See all 133

Metadata

Authors
Appel, M., & Prietzel, F.
Affiliations
University of Würzburg (DE)
Corresponding author
Appel, Markus
Venue
Journal of Computer-Mediated Communication
Published
2022-07-01
Volume
27
Issue
4
ISSN
1083-6101 (ISSN-L)
OpenAlex
W4285246761
Cited by
95
References
64
Citations by year
2026: 27 · 2025: 29 · 2024: 28 · 2023: 8
FWCI
23.38
Type
journal article
Language
en
Open access
Locations
Journal of Computer-Mediated Communication (publishedVersion, cc-by)
Topics
  • Social Sciences › Sociology and Political Science › Misinformation and Its Impacts
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Digital Media Forensic Detection
  • Social Sciences › Psychology › Social Psychology › Deception detection and forensic psychology
Keywords
deepfake detection, political deepfakes, fake news, analytical thinking, political interest, misinformation, manipulated video, perceived accuracy, Boris Johnson, Barack Obama, information processing model

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-06, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-06 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: The detection of political deepfakes. https://app.cochat.ai/papers/10.1093/jcmc/zmac008
The paper
Appel, M., & Prietzel, F. (2022). The detection of political deepfakes. Journal of Computer-Mediated Communication. https://doi.org/10.1093/jcmc/zmac008

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.