AI-synthesized faces are indistinguishable from real faces and more trustworthy
Open accessThe DOI matches its Crossref record; 18 of the 18 references it cites were found in a registry or an index.
Abstract
Artificial intelligence (AI)–synthesized text, audio, image, and video are being weaponized for the purposes of nonconsensual intimate imagery, financial fraud, and disinformation campaigns. Our evaluation of the photorealism of AI-synthesized faces indicates that synthesis engines have passed through the uncanny valley and are capable of creating faces that are indistinguishable—and more trustworthy—than real faces.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Figures




Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
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?
The cited source says what the claim attributes to it.
In the citing paper
“In the context of images, multiple studies show humans do not perform much better than chance [22, 23].”
From Warning: Humans cannot reliably detect speech deepfakes (2023) · cites this paper as [23]
The cited source says what the claim attributes to it.
In the citing paper
“Since early-generation LLMs, humans have been consistently poor at distinguishing AI-generated text (Mink et al., 2024), audio (Warren et al., 2024;Barrington et al., 2025) images (Nightingale & Farid, 2022;Mink et al., 2024), videos (Groh et al., 2022), and social media profiles containing AI-generated media (Mink et al., 2024).”
From Open Challenges in Multi-Agent Security: Towards Secure Systems of Interacting AI Agents (2025) · cites this paper as Nightingale & Farid, 2022
The cited source says what the claim attributes to it.
In the citing paper
“Images of synthetic faces produced by StyleGAN210 are indistinguishable from the original photos on which the StyleGAN2 algorithm was trained41.”
From Human detection of political speech deepfakes across transcripts, audio, and video (2024) · cites this paper as [41]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Analyzing and Improving the Image Quality of StyleGAN
Karras, T., Laine, S., Aittala, M., Hellsten, J., Lehtinen, J., & Aila, T. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr42600.2020.00813 · Cited by 6,332
Analyzing and Improving the Image Quality of StyleGAN
Karras, T., Laine, S., Aittala, M., Hellsten, J., Lehtinen, J., & Aila, T. (2019) · arXiv · doi:10.48550/arxiv.1912.04958 · Cited by 6,332 · cited 5×
“We hypothesize that White faces are more difficult to classify because they are overrepresented in the StyleGAN2 training dataset and are therefore more realistic.”
BackgroundMethodInfluentialParkhi, O. M., Vedaldi, A., & Zisserman, A. (2015) · Procedings of the British Machine Vision Conference 2015 · doi:10.5244/c.29.41 · Cited by 5,479
WaveNet: a generative model for raw audio
Oord, A. V. D., Dieleman, S., Zen, H., Simonyan, K., Vinyals, O., Graves, A., Kalchbrenner, N., Senior, A., & Kavukcuoglu, K. (2016) · arXiv · doi:10.48550/arxiv.1609.03499 · Cited by 4,911 · cited 1×
“From synthesizing speech in anyone’s voice (1) to synthesizing an image of a fictional person (2) and swapping one person’s identity with another or altering what they are saying in a video (3), AI-synthesized content holds the power to entertain but also deceive.”
BackgroundFirst Impressions: Making Up Your Mind After a 100-Ms Exposure to a Face
Willis, J., & Todorov, A. (2006) · Psychological Science · doi:10.1111/j.1467-9280.2006.01750.x · Cited by 2,300
Face X-Ray for More General Face Forgery Detection
Li, L., Bao, J., Zhang, T., Yang, H., Chen, D., Wen, F., & Guo, B. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr42600.2020.00505 · Cited by 1,211
CNN-Generated Images Are Surprisingly Easy to Spot… for Now
Wang, S.-Y., Wang, O., Zhang, R., Owens, A., & Efros, A. A. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr42600.2020.00872 · Cited by 1,210
Synthesizing Obama: learning lip sync from audio
Suwajanakorn, S., Seitz, S. M., & Kemelmacher-Shlizerman, I. (2017) · ACM Transactions on Graphics · doi:10.1145/3072959.3073640 · Cited by 1,057
Citron, D. K., & Franks, M. A. (2014) · Digital Commons at University of Maryland Carey Law (University of Maryland Francis King Carey School of Law) · Cited by 579
(2019) · Cited by 415
Self-citationArtificial Fingerprinting for Generative Models: Rooting Deepfake Attribution in Training Data
Yu, N., Skripniuk, V., Abdelnabi, S., & Fritz, M. (2021) · 2021 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.01418 · Cited by 195 · cited 1×
“Safeguards could include, for example, incorporating robust watermarks into the image and video synthesis networks that would provide a downstream mechanism for reliable identification (14).”
BackgroundWhat Is Typical Is Good: The Influence of Face Typicality on Perceived Trustworthiness
Sofer, C., Dotsch, R., Wigboldus, D. H. J., & Todorov, A. (2015) · Psychological Science · doi:10.1177/0956797614554955 · Cited by 167
The conceptual structure of face impressions
Stolier, R. M., Hehman, E., Keller, M., Walker, M., & Freeman, J. B. (2018) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1807222115 · Cited by 128 · cited 1×
“In this study, 223 participants rated the trustworthiness of 128 faces taken from the same set of 800 faces on a scale of 1 (very untrustworthy) to 7 (very trustworthy) (9).”
BackgroundWhy Are Deep Representations Good Perceptual Quality Features?
Tariq, T., Tursun, O. T., Kim, M., & Didyk, P. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-58542-6_27 · Cited by 30 · cited 1×
“perceptually meaningful (16) representation of each synthetic face.”
BackgroundDigital forensics in a post-truth age
Farid, H. (2018) · Forensic Science International · doi:10.1016/j.forsciint.2018.05.047 · Cited by 25 · cited 1×
“4–6), current techniques are not efficient or accurate enough to contend with the torrent of daily uploads (7).”
BackgroundSelf-citationFace X-ray for More General Face Forgery DetectionPreprint
Li, L., Bao, J., Zhang, T., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1912.13458 · Cited by 21
Detecting CNN-Generated Facial Images in Real-World Scenarios
Hulzebosch, N., Ibrahimi, S., & Worring, M. (2020) · arXiv · doi:10.48550/arxiv.2005.05632 · Cited by 18 · cited 1×
“This hyperphotorealism is consistent with recent findings (10, 11).”
ResultMore Real than Real: A Study on Human Visual Perception of Synthetic FacesPreprint
Lago, F., Pasquini, C., Böhme, R., Dumont, H., Goffaux, V., & Boato, G. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2106.07226 · Cited by 8 · cited 1×
“This hyperphotorealism is consistent with recent findings (10, 11).”
Result
Cited by
Generative AI Literacy: Twelve Defining Competencies
Annapureddy, R., Fornaroli, A., & Gatica-Perez, D. (2025) · Digital Government: Research and Practice · doi:10.1145/3685680 · Cited by 333
“Nonetheless, this competency differs from them as it relates more closely to the specific applications of generative AI, which may vary from case to case; moreover, although the ethical and legal contexts are important, the user should also consider the social and professional context, which may…”
The Creativity of Text-to-Image Generation
Oppenlaender, J. (2022) · Proceedings of the 25th International Academic Mindtrek Conference · doi:10.1145/3569219.3569352 · Cited by 290
“But it can be expected that we will, in the near future, no longer be able to tell text-toimage artworks apart from creations made by humans [37].”
BackgroundEvaluating Large Language Models in Generating Synthetic HCI Research Data: a Case Study
Hämäläinen, P., Tavast, M., & Kunnari, A. (2023) · Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3544548.3580688 · Cited by 248
Artificial intelligence and consumer behavior: From predictive to generative AI
Hermann, E., & Puntoni, S. (2024) · Journal of Business Research · doi:10.1016/j.jbusres.2024.114720 · Cited by 230
The power of generative marketing: Can generative AI create superhuman visual marketing content?
Hartmann, J., Exner, Y., & Domdey, S. (2025) · International Journal of Research in Marketing · doi:10.1016/j.ijresmar.2024.09.002 · Cited by 229
Metadata
- Authors
- Nightingale, S. J., & Farid, H.
- Affiliations
- Lancaster University (GB)
- University of California, Berkeley (US)
- Corresponding author
- Nightingale, Sophie J.
- Venue
- Proceedings of the National Academy of Sciences
- Publisher
- National Academy of Sciences
- Published
- 2022-02-14
- Volume
- 119
- Issue
- 8
- ISSN
- 0027-8424 (ISSN-L)
- 1091-6490
- PMC
- PMC8872790
- OpenAlex
- W4213187648
- Cited by
- 397
- References
- 13
- Citations by year
- 2026: 118 · 2025: 110 · 2024: 84 · 2023: 57 · 2022: 28
- FWCI
- 22.82
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- Proceedings of the National Academy of Sciences (publishedVersion, cc-by-nc-nd)
- PubMed Central (publishedVersion, cc-by-nc-nd)
- PubMed (publishedVersion)
- Lancaster EPrints (Lancaster University) (submittedVersion, cc-by-nc-nd)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Digital Media Forensic Detection
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Face recognition and analysis
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Generative Adversarial Networks and Image Synthesis
- Keywords
- AI-generated faces, face synthesis, photorealism, uncanny valley, face perception, trustworthiness, deepfakes, disinformation, financial fraud
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
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.