AI-generated faces fool most people but training can help spot deepfakes, new research reveals

See if you can spot an AI deepfake with our test

Psychologist Dr Clare Sutherland is holding up two large photos. One shows the face of an Australian academic leading an international research study; the other is an AI-generated deepfake. Artificial intelligence has become so adept at creating realistic images, it is increasingly hard to figure out what is real or not. But can people be trained to spot an image of a human that has actually been created by a machine? That's a question Sutherland, from the University of Aberdeen, and her Australian colleague have been examining. But before we reveal the answer, have a go at this test - and note down your score. If you found that tough, you are not alone. It used to be far easier to spot computer-generated visual creations - often used by fraudsters - because AI would make blunders, like adding an extra finger or something else that was obviously weird. But AI learns from its mistakes. "Training on visual artifacts, like looking for a sixth finger or odd earrings, has had limited success, partly because the AI is getting too good, and fraudsters may avoid using pictures with obvious flaws anyway," explained Prof Amy Dawel. She is the woman with shoulder-length hair in the picture being held by Sutherland. The man's image is the fake. Dawel is the director of the Australian National University Emotions and Faces Lab. She has been leading a team of researchers in Australia, Canada and the UK to find out if people can be trained to rumble the AI imposters. The answer, for now at least, is yes - but learning to spot an AI fake requires a more subtle approach. Sutherland is leading the UK-based research at the University of Aberdeen. She said they had noticed they were getting a feel for which faces were real or AI just by looking at them. "So we thought, OK, it would be really interesting to see if we could teach other people this too," she said. For the experiments a pool of thousands of AI-generated faces was created using an AI image tool called StyleGAN3, one of the most realistic face generators available. Participants were tested before and after being given training The researchers trained participants in the studies by drawing their attention to six perceptual qualities: * Symmetry - AI often fails to recreate the quirks that make us human - a slightly drooping eyelid or a lop-sided smile. "If it's too good to be true, it probably isn't." * Proportionality - A similar concept. Very large noses or protruding ears are not typical of deepfake images. * Attractiveness - "AI faces tend to look more attractive," explains Sutherland. "That one is more subjective, an aesthetic judgement, but AI often creates faces that are pleasant looking." * Distinctiveness - "That could be something like 'what would make a face stand out in a crowd?' AI faces do tend to cluster towards the average. So they look a bit more generic." * Expressiveness - "AI faces tend to look less emotionally expressive", says Sutherland. "They tend to show less emotion." * Memorability - "They often look less memorable - they're difficult to remember." AI also tends to be less proficient at recreating non-white, older or younger faces because more of its training involves young white people. Some of these tips might sound quite similar and "fuzzy" - but that's the point. Rarely will you encounter a surefire "tell" that will unmask an AI fake. Rather, it is about becoming attuned to their characteristics and developing a gut feeling. Researchers found that by exposing people to images, both AI and real, then telling them which was which, they can get significantly better at it - even in the space of an hour or so. The researchers found the participants would typically increase their accuracy score from about 40% to 80%. A few individuals achieved close to 100% accuracy. Ironically, what the human brain is doing here is similar to the way that generative AI models work. Give them enough data to train on and, over time, their accuracy improves - even though we may not totally understand how they are doing it. The studies also looked at how confident the participants were at identifying the AI images. Previous research had indicated people were overconfident that they could spot AI faces, with the most confident people making the most errors. After training, participants were found to have increased their confidence in spotting the deepfakes. "That's helpful right?" says Sutherland. "Because if you don't know when you're correct or not, you can't really do anything with that information." OK, so are you ready to take another test? How did you do? Feeling more confident? If the answer is no, don't beat yourself up over it. In both the human world and that of generative AI, practice makes perfect - or at least a bit closer to perfect. There are many websites out there where, if you so desire, you can hone your skills. You can also volunteer to take part in the research yourself. The obvious danger is fraud. Global consultancy firm Deloitte has predicted that losses from AI deepfake scams in the US alone could rise to £40bn next year, up from £12bn in 2023. The report cited the example of a scam where an employee at a Hong Kong-based firm transferred £25m to fraudsters after a video call with a deepfake recreation of their boss. Another sinister use of deepfake technology is political espionage As long ago as 2019, an Associated Press investigation found that a LinkedIn profile - including a photo - belonging to a woman called Katie Jones appeared to be fictitious. Jones purported to be a Russia and Eurasia specialist with links to prominent Washington think tanks and policy circles. The AP report claimed she was actually a deepfake produced by Russian intelligence who had successfully connected with top US political aides and national security officials. In Australia, a politician is currently proposing a requirement to disclose and "watermark" AI-generated political content. To be fair to AI, Sutherland also sees some positive uses of the technology - such as the ability to quickly and cheaply show how a long-missing child might look at various ages. She says that if people are "engaging with it in good faith and people know that AI has been used, it could potentially be very useful for creative acts". So the good news is that we're yet not living in a dystopian world where it's impossible to tell what's real and what's computer-generated. The bad news is that AI models may have already "read" the published academic research papers. And it's learning.

Study finds AI-generated faces seem more trustworthy than real people

People trust faces created by artificial intelligence more than they trust photos of actual human beings, according to new research. The findings raise fresh alarm bells about online fraud, catfishing, and misinformation. The finding comes at a time when AI image generators have become accessible to virtually anyone, regardless of technical skill. That has put a powerful new tool for deception within reach of ordinary internet users rather than just specialists. The study, led by Alexis McGuire and a team from Lancaster University, is the first to examine how trustworthy people find AI faces created using the latest diffusion technology. This newer approach to image generation is more sophisticated than earlier AI systems. Spotting fake faces isn't easy Human beings are remarkably good at reading real faces, capable of forming judgements in as little as 100 milliseconds, a skill built up over the entire span of human evolution. But AI-generated faces have become so realistic that newer, more sophisticated systems can now fool people into thinking a fake face is genuine roughly a third of the time. That erodes an ability that once felt almost instinctive. To test this directly, researchers showed 169 participants a set of 96 faces spanning a range of races, genders, and ages. Researchers presented the faces in random order, and participants judged whether each one was real or AI-generated. The average accuracy came out to just 58.4%, only marginally better than flipping a coin. Oddly enough, participants rated faces from the newer diffusion model as less realistic than those from an older AI system known as a GAN. That was despite the diffusion model representing more advanced underlying technology. That result alone hinted that something more complicated than simple image quality was shaping people's judgments. AI faces win our trust A follow-up experiment pushed the question further. Researchers asked a new group of participants to rate the trustworthiness of 96 faces on a scale from one, very untrustworthy, to seven, very trustworthy. This approach separates the question of trust entirely from the earlier question of realism. Real human faces came out lowest, with an average trust score of 4.03. Both types of AI-generated faces scored higher across the board. GAN-produced faces averaged 4.36, while faces from the newer diffusion model scored highest of all, at 4.70. That means the AI faces people found least realistic in the first experiment were, somewhat bizarrely, the very ones they trusted most in the second. "This finding presents a paradox and thus highlights the possibility that realism and trustworthiness judgements are driven by two different psychological mechanisms," McGuire said. In other words, whatever process the brain uses to size up how believable a face looks may not be the same process it uses to decide whether that face seems safe or friendly. How scammers could benefit "Our research shows that people are at risk of being fooled by AI-generated images," McGuire said. "These AI models have democratized the online space, and they are accessible to anyone without technical skills who wants to create fake faces that can be used for a variety of harms." McGuire stressed the importance of educating the public about how easily people can generate convincing AI faces and the risks they pose, including misinformation, identity fraud, and catfishing. She also pointed to a broader concern that extends well past individual scams and stretches toward institutions and public life more generally. "As AI-generated images become more sophisticated and more accessible, as a society, we are increasingly exposed to AI-generated faces, often in nefarious and exploitative scenarios," said McGuire. "It is critical to understand the threat this democratization of generative AI brings, as well as to develop strategies to mitigate potential harms to individuals, organisations, and democracies." A problem with no easy fix The findings suggest that as diffusion models continue improving, the gap between how trustworthy AI faces seem and how trustworthy they actually are could keep widening. That gap gives scammers, catfishers, and disinformation campaigns a powerful opportunity to deceive people. What makes this especially concerning is the direction the technology seems to be heading. If newer AI models keep producing faces that people rate as increasingly trustworthy, simple visual instinct may become an increasingly unreliable defense against fabricated identities online. That could happen even if those faces do not necessarily look more realistic. Can you spot AI faces? The research team is continuing to study how people process real versus AI-generated faces. Anyone interested can take part in an ongoing anonymous online survey called "Examining Individual Differences in the Detection of Real and AI-generated Faces." Participants view a series of faces, rate whether each looks real or AI-generated along with their confidence level, and receive a score at the end reflecting how well they did. This will contribute significantly to a growing body of evidence about just how vulnerable human judgment has become in the face of increasingly convincing synthetic imagery. The study is published in the Journal of Vision. -- - Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates. Check us out on EarthSnap, a free app brought to you by Eric Ralls and Earth.com.