CoChat
Sign Up
Cited by 33Misinformation and Its ImpactsDigital Media Forensic DetectionEthics and Social Impacts of AI

Deepfakes: A Grounded Threat Assessment

Open access
Center for Security and Emerging Technology, Hwang, Tim·2020
DOI 10.51593/20190030
Open this paper in CoChat — free accountOpen PDF

Abstract

The rise of deepfakes could enhance the effectiveness of disinformation efforts by states, political parties and adversarial actors. How rapidly is this technology advancing, and who in reality might adopt it for malicious ends? This report offers a comprehensive deepfake threat assessment grounded in the latest machine learning research on generative models.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.51593/20190030

Cites

5 of 5 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

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?

  • Existing unimodal deepfake detectors are primarily designed to detect a single type of manipulation, such as visual, acoustic, and text.

    “Existing unimodal deepfake detectors are primarily designed to detect a single type of manipulation, such as visual, acoustic, and text.”

    From Contextual Cross-Modal Attention for Audio-Visual Deepfake Detection and Localization (2024) · cites this paper as [24]

    The “Deepfake Detection: Strengths and Weaknesses” section describes detectors targeting visual artifacts and says systems that identify known creation methods may generalize poorly to others. It does not establish that existing detectors are primarily unimodal or describe detectors separately designed for acoustic and text manipulations; the Introduction says text generation is outside the paper’s review. (LLM, conf=0.78)

    Experimental — accuracy not yet measured

  • This preprocessing increases the pixel level statistical similarity between real images and fake images and allows the forensic classifier to learn more intrinsic and meaningful features, which has better generalization capability than previous image forensics methods or image steganalysis networks.

    “This increases the pixel level statistical similar- ity between real images and fake images and allows the forensic classifier to learn more intrinsic and meaning- ful features, which has better generalization capability than previous image forensics methods [75, 76] or im- age steganalysis networks [77].”

    From Deep learning for deepfakes creation and detection: A survey (2022) · cites this paper as [77]

    Split into 2 atoms — 0 confirmed, 0 partially supported, 0 disputed, 0 misattributed, 2 unverifiable. Expand the claim to see per-atom verdicts.

    Experimental — accuracy not yet measured

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

Flagged first, then by citations

Research works printed but not linked: 33

  1. Shane Barratt and Rishi Sharma, "A Note on the Inception Score." Preprint, submitted January 6, 2018, https://arxiv.org/abs/1801.01973; · arXiv
  2. Konstantin Shmelkov, Cordelia Schmid, and Karteek Alahair, "How good is my GAN?" Preprint, submitted July 25, 2018, https://arxiv.org/ abs/1807.09499. · arXiv
  3. Mario Lucic et al., "Are GANs Created Equal? A Large-Scale Study." Preprint, submitted November 28, 2017, http://arxiv.org/abs/1711.10337. · arXiv
  4. Diederik Kingma and Prafulla Dhariwal, "Glow: Generative Flow with Invertible 1x1 Convolutions." Preprint, submitted July 9, 2018. https://arxiv.org/abs/1807.03039 (Glow); · arXiv
  5. Aaron van den Lord, Nal Kalchbrenner, and Koray Kavukcuoglu, "Pixel Recurrent Neural Networks." Preprint, submitted January 25, 2016, https://arxiv.org/abs/1601.06759 (PixelCNN); · arXiv
  6. Benigno Uria et al., "Neural Autoregressive Distribution Estimation." Preprint, submitted May 7, 2016, https://arxiv.org/ abs/1605.02226 (NADE); · arXiv
  7. Karol Gregor et al., "DRAW: A Recurrent Neural Network for Image Generation." Preprint, submitted May 20, 2015, http://arxiv.org/abs/1502.04623 (DRAW). · arXiv
  8. Aaron van den Oord et al., "WaveNet: A Generative Model for Raw Audio." Preprint, submitted September 19, 2016, http://arxiv.org/abs/1609.03499. · arXiv
  9. Carl Vondrick, Hamed Pirsiavash, and Antonio Torralba, "Generating Videos with Scene Dynamics." Preprint, submitted September 8, 2016, https://arxiv.org/abs/1609.02612. · arXiv
  10. For more detail on this process, see Ian Goodfellow, Yoshua Bengio, and Aaron Courville, Deep Learning (Cambridge: MIT Press, 2016), Chapter 20.10.4, https://www.deeplearningbook.org/ contents/generative_models.html.
  11. Augustus Odena, Christopher Olah, and Jonathon Shlens, "Conditional Image Synthesis With Auxiliary Classifier GANs." Preprint, submitted October 30, 2016, https://arxiv.org/ abs/1610.09585; · arXiv
  12. Marco Marchesi, "Megapixel Size Image Creation using Generative Adversarial Networks." Preprint, submitted March 31, 2017, https://arxiv.org/abs/1706.00082; · arXiv
  13. David Berthelot, Thomas Schumm, and Luke Metz, "BEGAN: Boundary Equilibrium Generative Adversarial Networks." Preprint, March 31, 2017, https://arxiv.org/abs/1703.10717. · arXiv
  14. Ming-Yu Liu, Thomas Breuel, and Jan Kautz, "Unsupervised Image-to-Image Translation Networks." Preprint, submitted March 2, 2017, http://arxiv.org/abs/1703.00848. · arXiv
  15. Grigory Antipode, Moez Baccouche, and Jean-Luc Dugelay, "Face Aging With Conditional Generative Adversarial Networks." Preprint, submitted February 7, 2017, https://arxiv.org/ abs/1702.01983. · arXiv
  16. Jun-Yan Zhu et al., "Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks." Preprint, submitted March 30, 2017, http://arxiv.org/abs/1703.10593. · arXiv
  17. Tero Karras et al., "Progressive Growing of GANs for Improved Quality, Stability, and Variation." Preprint, submitted October 27, 2017, https://arxiv.org/abs/1710.10196; · arXiv
  18. Tero Karras, Samuli Laine, and Timo Aila, "A Style-Based Generator Architecture for Generative Adversarial Networks." Preprint, submitted December 12, 2018, https://arxiv.org/abs/1812.04948 (a more recent paper showing a similar training arrangement). · arXiv
  19. van den Oord et al., "Wavenet"; Yuxuan Wang et al., "Tacotron: Towards End-to-End Speech Synthesis." Preprint, submitted March 29, 2017, http://arxiv.org/abs/1703.10135. · arXiv
  20. Soroush Mehri, "SampleRNN: An Unconditional End-to-End Neural Audio Generation Model." Preprint, submitted February 11, 2017, http://arxiv.org/abs/1612.07837. · arXiv
  21. Sercan O. Arik et al., "Deep Voice: Real-time Neural Text-to-Speech." Preprint, submitted February 25, 2017, http://arxiv.org/abs/1702.07825. · arXiv
  22. For a brief review of recent literature on these techniques, see Caroline Chan et al., "Everybody Dance Now." Preprint, submitted August 22, 2018, http://arxiv.org/abs/1808.07371. · arXiv
  23. Chan et al., "Everybody." 44. Jessica Lee, Deva Ramanan and Rohit Girdhar, "MetaPix: Few-Shot Video Retargeting." Preprint, submitted October 10, 2019, http://arxiv.org/abs/1910.04742. · arXiv
  24. Two useful surveys reviewing techniques in this field include Raahat Devender Singh and Naveen Aggarwal, "Video content authentication techniques: a comprehensive survey," Multimedia Systems 24 (2018): 221-240, https://link.springer.com/article/10.1007/s00530-017-0538-9 and Luisa Verdoliva, "Media Forensics and DeepFakes: an overview." Preprint, submitted January 18, 2020, https://arxiv.org/abs/2001.06564. · DOI
  25. Yuezun Li, Ming-Ching Chang, and Siwei Lyu, "In Ictu Oculi: Exposing AI Generated Fake Face Videos by Detecting Eye Blinking." Preprint, submitted June 7, 2018, https://arxiv.org/abs/1806.02877. · arXiv
  26. Andreas Rossler et al., "FaceForensics++: Learning to Detect Manipulated Facial Images." Preprint, submitted January 25, 2019, https://arxiv.org/abs/1901.08971. · arXiv
  27. Xin Yang et al., "Exposing GAN-synthesized Faces Using Landmark Locations." Preprint, submitted March 30, 2019, http://arxiv.org/abs/1904.00167; · arXiv
  28. Yuezun Li and Siwei Lyu, "Exposing DeepFake Videos By Detecting Face Warping Artifacts." Preprint, submitted November 1, 2018, http://arxiv. org/abs/1811.00656.
  29. Ali Khodabakhsh et al., "Fake Face Detection Methods: Can They Be Generalized?" Paper presented at the 2018 International Conference of the Biometrics Special Interest Group (BIOSIG), Darmstadt, Germany, September 2018, https://ieeexplore.ieee.org/document/8553251/.
  30. Gordon Pennycook et al., "The Implied Truth Effect: Attaching Warnings to a Subset of Fake News Headlines Increases Perceived Accuracy of Headlines Without Warnings." Preprint, submitted September 14, 2017, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3035384.
  31. Sharad Goal, Duncan Watts, and Daniel Goldstein, "The structure of online diffusion networks," Proceedings of the 13th ACM Conference on Electronic Commerce (2012): 623-638, https://dl.acm. org/doi/10.1145/2229012.2229058. · DOI
  32. Daniela Flamini, "WhatsApp efforts to curb misinformation aren't entirely effective, research shows," Poynter, September 27, 2019, https://www.poynter.org/fact-checking/2019/whatsapp-efforts-to- curb-misinformation-arent-entirely-effective-research-shows/; Sharon Moshavi, "Op-ed: Shining light into the dark spaces of chat apps," Columbia Journalism Review, January 14, 2020, https://www.cjr. org/opinion/whatsapp-messenger-misinformation.php.
  33. For a sample of the vast literature on these topics, see Joshua Tucker et al., Social Media, Political Polarization, and Political Disinformation: A Review of the Scientific Literature (Menlo Park, CA:

Cited by

Sorted by citations

Authors
Center for Security and Emerging Technology, & Hwang, T.
Corresponding author
Hwang, Tim
Publisher
Center for Security and Emerging Technology
Published
2020-07
DOI
10.51593/20190030
OpenAlex
W3128683850
Cited by
33
References
5
Citations by year
2026: 5 · 2025: 2 · 2024: 7 · 2023: 9 · 2022: 3 · 2021: 7
Type
report
Language
en
Open access
Licence
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 › Safety Research › Ethics and Social Impacts of AI
Keywords
deepfakes, disinformation, threat assessment, generative models, machine learning, state-sponsored disinformation, political manipulation, artificial intelligence adoption

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.

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