What Happened
Researchers at the ANU Emotions and Faces Lab published findings on 9 July 2026 showing that targeted training significantly improves people's ability to identify AI-generated faces. The study tested participants on StyleGAN faces, a class of synthetic images considered among the most convincing currently produced.
The training method moves away from teaching people to look for obvious errors. As Associate Professor Amy Dawel explained: "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."
Instead, the program directs attention to broader face-level qualities. "Our training directs people's attention to global qualities that differ between AI and human faces. AI faces tend to be more symmetrical, proportional and attractive, but without training we often think these are markers of being human," Associate Professor Dawel said.
The results were replicated with participants in a second country. Dr Mah noted: "The replication shows that the findings weren't a fluke. When we trained a new set of people in a different country, we saw them improve just as much. Online training was effective, so our training program could easily be implemented at scale for little cost."
Why It Matters
Deepfake fraud is a growing concern in Australia and internationally. AI-generated faces are now realistic enough that most people cannot reliably distinguish them from genuine photographs, creating real risks in identity verification, romance scams, and synthetic media.
Associate Professor Dawel was direct about the limits of relying solely on automated tools: "While algorithms offer one solution to detecting deepfake faces, their decision-making processes remain opaque and recent benchmarking reveals serious weaknesses. We need approaches that are ethical and explainable, for which keeping humans in the loop is key."
Key Details
The ANU study targeted StyleGAN faces specifically because they represent a high bar for detection. Associate Professor Dawel said: "We've shown our training is effective for some of the most convincing fakes available, StyleGAN faces. Now we need to find out whether that training generalises to other AI-generated faces."
The online delivery format is central to the program's practical value. Because training ran effectively over the internet, the researchers say it could reach large numbers of people without significant infrastructure costs.
Background and Context
AI image generation has advanced rapidly, and public awareness has not kept pace. As one researcher noted: "AI image-generation technology is improving extremely quickly, and many people underestimate how convincing these faces can be. Research like this can help people navigate increasingly complex online environments."
The ANU Emotions and Faces Lab has focused on the perceptual science of face recognition, applying that work to the practical problem of synthetic media detection. The deepfake detection challenge sits at the intersection of cognitive psychology and AI safety, a combination that distinguishes this approach from purely technical solutions.
What Comes Next
The research team plans to test whether the training generalises beyond StyleGAN to other AI face-generation systems. If it does, the program could form the basis of a broadly deployable public awareness tool. The low cost of online delivery makes integration into workplace fraud-prevention training or consumer education programs feasible without major funding.