Content uploaded by Sophie Nightingale
Author content
All content in this area was uploaded by Sophie Nightingale on Feb 23, 2022
Content may be subject to copyright.
BRIEF REPORT
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
AI-synthesized faces are indistinguishable from real
faces and more trustworthy
Sophie J. Nightingalea,1 and Hany Faridb
aDepartment of Psychology, Lancaster University, Lancaster LA1 4YW, United Kingdom; and bDepartment of Electrical Engineering and Computer Sciences,
University of California, Berkeley, CA 94720
Edited by William Press, Computer Sciences and Integrative Biology, University of Texas at Austin, Austin, TX; received November 11, 2021; accepted
December 20, 2021
Articial intelligence (AI)–synthesized text, audio, image, and
video are being weaponized for the purposes of nonconsensual
intimate imagery, nancial fraud, and disinformation campaigns.
Our evaluation of the photorealism of AI-synthesized faces indi-
cates that synthesis engines have passed through the uncanny val-
ley and are capable of creating faces that are indistinguishable—
and more trustworthy—than real faces.
deep fakes | face perception
Artificial intelligence (AI)–powered audio, image, and video
synthesis—so-called deep fakes—has democratized access
to previously exclusive Hollywood-grade, special effects technol-
ogy. From synthesizing speech in anyone’s voice (1) to synthesiz-
ing 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.
Generative adversarial networks (GANs) are popular mech-
anisms for synthesizing content. A GAN pits two neural
networks—a generator and discriminator—against each other.
To synthesize an image of a fictional person, the generator starts
with a random array of pixels and iteratively learns to synthesize
a realistic face. On each iteration, the discriminator learns to
distinguish the synthesized face from a corpus of real faces; if the
synthesized face is distinguishable from the real faces, then the
discriminator penalizes the generator. Over multiple iterations,
the generator learns to synthesize increasingly more realistic
faces until the discriminator is unable to distinguish it from real
faces (see Fig. 1 for example real and synthetic faces).
Much has been written in the popular press about the potential
threats of deep fakes, including the creation of nonconsensual
intimate imagery (more commonly referred to by the misnomer
“revenge porn”), small- to large-scale fraud, and adding jet fuel
to already dangerous disinformation campaigns. Perhaps most
pernicious is the consequence that, in a digital world in which any
image or video can be faked, the authenticity of any inconvenient
or unwelcome recording can be called into question.
Although progress has been made in developing automatic
techniques to detect deep-fake content (e.g., refs. 4–6), current
techniques are not efficient or accurate enough to contend with
the torrent of daily uploads (7). The average consumer of online
content, therefore, must contend with sorting out the real from
the fake. We performed a series of perceptual studies to deter-
mine whether human participants can distinguish state-of-the-art
GAN-synthesized faces from real faces and what level of trust the
faces evoked.
Results
Experiment 1. In this study, 315 participants classified, one at
a time, 128 of the 800 faces as real or synthesized. Shown in
Fig. 2Ais the distribution of participant accuracy (blue bars).
The average accuracy is 48.2% (95% CI [47.1%, 49.2%]),
close to chance performance of 50%, with no response bias:
d=−0.09;β=0.99. Two repeated-measures binary logistic
regression analyses were conducted—one for real and one for
synthetic faces—to examine the effect of stimuli gender and race
on accuracy. For real faces, there was a significant gender ×
race interaction, χ2(3, N= 315) = 95.03, P<0.001. Post hoc
Bonferroni-corrected comparisons revealed that mean accuracy
was higher for male East Asian faces than female East Asian
faces and higher for male White faces than female White faces.
For synthetic faces, there was also a significant gender ×race
interaction, χ2(3, N= 315) = 68.41, P<0.001. For both male
and female synthetic faces, White faces were the least accurately
classified, and male White faces were less accurately classified
than female White faces. We hypothesize that White faces are
more difficult to classify because they are overrepresented in the
StyleGAN2 training dataset and are therefore more realistic.
Experiment 2. In this study, 219 new participants, with training
and trial-by-trial feedback, classified 128 faces taken from the
same 800 set of faces as in experiment 1. Shown in Fig. 2A
is the distribution of participant accuracy (orange bars). The
average accuracy improved slightly to 59.0% (95% CI [57.7%,
60.4%]), with no response bias: d=0.46;β=0.99.Despite
providing trial-by-trial feedback, there was no improvement in
accuracy over time, with an average accuracy of 59.3% (95% CI
[57.8%, 60.7%]) for the first set of 64 faces and 58.8% (95% CI
[57.4%, 60.3%]) for the second set of 64 faces. Further analyses
to examine the effect of gender and race on accuracy replicated
the primary findings of experiment 1. This analysis again revealed
that, for both male and female synthetic faces, White faces were
the most difficult to classify.
When made aware of rendering artifacts and given feedback,
there was a reliable improvement in accuracy; however, overall
performance remained only slightly above chance. The lack of
improvement over time suggests that the impact of feedback is
limited, presumably because some synthetic faces simply do not
contain perceptually detectable artifacts.
Experiment 3. Faces provide a rich source of information, with
exposure of just milliseconds sufficient to make implicit infer-
ences about individual traits such as trustworthiness (8). We
wondered whether synthetic faces activate the same judgements
of trustworthiness. If not, then a perception of trustworthiness
could help distinguish real from synthetic faces.
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). Shown in Fig. 2Bis the
distribution of average ratings (by averaging the ordinal ratings,
we are assuming a linear rating scale). The average rating for real
Author contributions: S.J.N. and H.F. designed research, performed research, contributed
new reagents/analytic tools, analyzed data, and wrote the paper.
The authors declare no competing interest.
This open access article is distributed under Creative Commons Attribution-
NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND).
1To whom correspondence may be addressed. Email: s.nightingale1@lancaster.ac.uk.
Published February 14, 2022.
PNAS 2022 Vol. 119 No. 8 e2120481119 https://doi.org/10.1073/pnas.2120481119 1of3
Downloaded by guest on February 17, 2022
Fig. 1. The most (Top and Upper Middle) and least (Bottom and Lower
Middle) accurately classied real (R) and synthetic (S) faces.
faces (blue bars) of 4.48 is less than the rating of 4.82 for synthetic
faces (orange bars). Although only 7.7% more trustworthy, this
difference is significant [t(222) = 14.6, P<0.001, d=0.49]. Al-
though a small effect, Black faces were rated more trustworthy
than South Asian faces, but, otherwise, there was no effect
across race. Women were rated as significantly more trustworthy
than men, 4.94 as compared to 4.36 [t(222) = 19.5, P<0.001,
d=0.82].
Shown in Fig. 3 are the four most (Fig. 3, To p) and four
least (Fig. 3, Bottom) trustworthy faces. The top three most
trustworthy faces are synthetic (S), while the bottom four least
trustworthy faces are real (R). A smiling face is more likely to
be rated as trustworthy, but 65.5% of our real faces and 58.8%
of synthetic faces are smiling, so facial expression alone cannot
explain why synthetic faces are rated as more trustworthy.
Discussion
Synthetically generated faces are not just highly photorealistic,
they are nearly indistinguishable from real faces and are judged
more trustworthy. This hyperphotorealism is consistent with re-
cent findings (10, 11). These two studies did not contain the
same diversity of race and gender as ours, nor did they match
the real and synthetic faces as we did to minimize the chance of
inadvertent cues. While it is less surprising that White male faces
are highly realistic—because these faces dominate the neural
network training—we find that the realism of synthetic faces
extends across race and gender. Perhaps most interestingly, we
find that synthetically generated faces are more trustworthy than
real faces. This may be because synthesized faces tend to look
more like average faces which themselves are deemed more
trustworthy (12). Regardless of the underlying reason, synthet-
ically generated faces have emerged on the other side of the
uncanny valley. This should be considered a success for the fields
of computer graphics and vision. At the same time, easy access
Fig. 2. The distribution of participant accuracy for (A) experiment 1 and
experiment 2 (chance performance is 50%), and (B) trustworthy ratings for
experiment 3 (a rating of 1 corresponds to the lowest trust).
(https://thispersondoesnotexist.com) to such high-quality fake
imagery has led and will continue to lead to various problems,
including more convincing online fake profiles and—as synthetic
audio and video generation continues to improve—problems of
nonconsensual intimate imagery (13), fraud, and disinformation
campaigns, with serious implications for individuals, societies,
and democracies.
We, therefore, encourage those developing these technologies
to consider whether the associated risks are greater than their
benefits. If so, then we discourage the development of technology
simply because it is possible. If not, then we encourage the par-
allel development of reasonable safeguards to help mitigate the
inevitable harms from the resulting synthetic media. 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). Because
it is the democratization of access to this powerful technology
that poses the most significant threat, we also encourage recon-
sideration of the often laissez-faire approach to the public and
unrestricted releasing of code for anyone to incorporate into any
application.
Fig. 3. The four most (Top ) and four least (Bottom) trustworthy faces and
their trustworthy rating on a scale of 1 (very untrustworthy) to 7 (very
trustworthy). Synthetic faces (S) are, on average, more trustworthy than real
faces (R).
2of3 PNAS
https://doi.org/10.1073/pnas.2120481119
Nightingale and Farid
AI-synthesized faces are indistinguishable from
real faces and more trustworthy
Downloaded by guest on February 17, 2022
BRIEF REPORT
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
Fig. 4. A representative set of matched real and synthetic faces.
At this pivotal moment, and as other scientific and engineering
fields have done, we encourage the graphics and vision commu-
nity to develop guidelines for the creation and distribution of
synthetic media technologies that incorporate ethical guidelines
for researchers, publishers, and media distributors.
Materials and Methods
Synthetic Faces. We selected 400 faces synthesized using the state-of-the-
art StyleGAN2 (2), ensuring diversity across gender (200 women; 200 men),
estimated age (ensuring a range of ages from children to older adults), and
race (100 African American or Black, 100 Caucasian, 100 East Asian, and 100
South Asian). To reduce extraneous cues, we only included images with a
mostly uniform background, and devoid of any obvious rendering artifacts.
This culling of obvious artifacts makes the perceptual task harder. Because
the synthesis process is so easy, however, it is reasonable to assume that any
intentionally deceptive use of a synthetic face will not contain obvious visual
artifacts.
Real Faces. For each synthesized face, we collected a matching real face (in
terms of gender,age, race, and overall appearance) from the underlying face
database used in the StyleGAN2 learning stage. A standard convolutional
neural network descriptor (15) was used to extract a low-dimensional,
perceptually meaningful (16) representation of each synthetic face. The
extracted representation for each synthetic face—a 4,096-D real-valued
vector
vs—was compared with all other facial representations in the data
set of 70,000 real faces to nd the most similar face. The real face with
representation
vrwith minimal Euclidean distance to
vs, and satisfying our
qualitative selection criteria, is selected as the matching face. As with the
synthetic faces, to reduce extraneous cues, we only included images 1) with
a mostly uniform background, 2) with unobstructed faces (e.g., no hats or
hands in front of face), 3) in focus and high resolution, and 4) with no
obvious writing or logos on clothing. We visually inspected up to 50 of the
best matched faces and selected the one that met the above criteria and was
also matched in terms of overall face position, posture, and expression, and
presence of glasses and jewelry. Shown in Fig. 4 are representative examples
of these matched real and synthetic faces.
Perceptual Ratings. For experiment 1 (baseline), we recruited 315 partici-
pants from Amazon’s Mechanical Turk Master Workers. Each participant rst
read a brief introduction explaining the purpose of the study and a brief
explanation of what a synthetic face is. Before beginning, each participant
was informed they would be paid $5 for their time, and an extra $5 if their
overall accuracy was in the top 20% of response accuracies. Participants were
also informed they would see 10 catch trials of obviously synthetic faces
with glaring rendering errors. A failure to respond correctly to at least nine
of these trials led to the participants not being paid and their data being
excluded from our study. Each participant then saw 128 images, one at a
time, and specied whether the image was real or synthetic. Participants
had an unlimited amount of time to respond and were not provided with
feedback after each response.
For experiment 2 (training and feedback), we recruited another 219
Mechanical Turk Master Workers (we had fewer participants in this study
because we excluded any participants who completed the rst study). Each
participant rst read a brief introduction explaining the purpose of the study
and a brief explanation of what a synthetic face is. Participants were then
shown a short tutorial describing examples of specic rendering artifacts
that can be used to identify synthetic faces. All other experimental condi-
tions were the same as in experiment 1, except that participants received
feedback after each response.
For experiment 3 (trustworthiness), we recruited 223 Mechanical Turk
Master Workers. Each participant rst read a brief introduction explaining
that the purpose of the study was to assess the trustworthiness of a face on
a scale of 1 (very untrustworthy) to 7 (very trustworthy). Because there was
no correct answer here, no trial-by-trial feedback was provided. Participants
were also informed they would see 10 catch trials of faces in which the
numeric trustworthy rating was directly overlaid atop the face. A failure
to correctly report the specied rating on at least nine of these trials led
to the participants not being paid and their data being excluded from our
study. Each participant then saw 128 images, one at a time, and was asked
to rate the trustworthiness. Participants had an unlimited amount of time
to respond.
All experiments were carried out with the approval of the University of
California, Berkeley’s Ofce for Protection of Human Subjects (Protocol ID
2019-07-12422) and Lancaster University’s Faculty of Science and Technology
Research Ethics Committee (Protocol ID FST20076). Participants gave fully
informed consent prior to taking part in the study.
Data Availability. Images have been deposited in GitHub (https://
github.com/NVlabs/stylegan2 and https://github.com/NVlabs/ffhq-dataset).
Anonymized experimental stimuli and data have been deposited in the
Open Science Framework (https://osf.io/ru36d/).
ACKNOWLEDGMENTS. We thank Erik Härkönen, Jaakko Lehtinen, and
David Luebke for their masterful synthesis of faces.
1. A. Oord et al., WaveNet: A generative model for raw audio. arXiv [Preprint] (2016).
https://arxiv.org/abs/1609.03499 (Accessed 17 January 2022).
2. T. Karras et al., “Analyzing and improving the image quality of StyleGAN” in IEEE
Conference on Computer Vision and Pattern Recognition (Institute of Electrical and
Electronics Engineers, 2020), pp. 8110–8119.
3. S. Suwajanakorn, S. M. Seitz, I. Kemelmacher-Shlizerman, Synthesizing Obama:
Learning lip sync from audio. ACM Trans. Graph. 36, 95 (2017).
4. L. Li et al., Face X-ray for more general face forgery detection. arXiv [Preprint]
(2019). https://arxiv.org/abs/1912.13458 (Accessed 17 January 2022).
5. S. Agarwal et al., “Protecting world leaders against deep fakes” in Proceedings of
the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops
(Institute of Electrical and Electronics Engineers, 2019), pp. 38–45.
6. S. Y. Wang, O. Wang, R. Zhang, A. Owens, A. A. Efros, “CNN-generated images
are surprisingly easy to spot... for now” in IEEE Conference on Computer Vision
and Pattern Recognition (Institute of Electrical and Electronics Engineers, 2020), pp.
8695–8704.
7. H. Farid, Commentary: Digital forensics in a post-truth age. Forensic Sci. Int. 289,
268–269 (2018).
8. J. Willis, A. Todorov, First impressions: Making up your mind after a 100-ms exposure
to a face. Psychol. Sci. 17, 592–598 (2006).
9. R. M. Stolier, E. Hehman, M. D. Keller, M. Walker, J. B. Freeman, The conceptual
structure of face impressions. Proc.Natl.Acad.Sci.U.S.A.115, 9210–9215 (2018).
10. F. Lago et al., More real than real: A study on human visual perception of synthetic
faces. arXiv [Preprint] (2021). https://arxiv.org/abs/2106.07226 (Accessed 17 January
2022).
11. N. Hulzebosch, S. Ibrahimi, M. Worring, “Detecting CNN-generated facial images
in real-world scenarios” in Proceedings of the IEEE/CVF Conference on Computer
Vision and Pattern Recognition Workshops (Institute of Electrical and Electronics
Engineers, 2020), pp. 642–643.
12. C. Sofer, R. Dotsch, D. H. Wigboldus, A. Todorov, What is typical is good: The
inuence of face typicality on perceived trustworthiness. Psychol. Sci. 26, 39–47
(2015).
13. D. K. Citron, M. A. Franks, Criminalizing revenge porn. Wake For. Law Rev. 49, 345
(2014).
14. N. Yu, V. Skripniuk, S. Abdelnabi, M. Fritz, “Articial ngerprinting for generative
models: Rooting deepfake attribution in training data” in Proceedings of the
IEEE/CVF International Conference on Computer Vision (Institute of Electrical and
Electronics Engineers, 2021), pp. 14448–14457.
15. O. M. Parkhi, A. Vedaldi, A. Zisserman, Deep Face Recognition (British Machine
Vision Association, 2015).
16. T. Tariq, O. T. Tursun, M. Kim, P. Didyk, “Why are deep representations good per-
ceptual quality features?” in European Conference on Computer Vision (Springer,
2020), pp. 445–461.
Nightingale and Farid
AI-synthesized faces are indistinguishable from
real faces and more trustworthy
PNAS 3of3
https://doi.org/10.1073/pnas.2120481119
Downloaded by guest on February 17, 2022
