A research team from Swansea University, the University of Lincoln, and Ariel University in Israel has found that artificial intelligence can now generate images of people’s faces—both fictional and real—that are almost impossible for viewers to distinguish from authentic photographs.
The study used AI models ChatGPT and DALL·E to create realistic facial images, including those of well-known celebrities. The researchers conducted four experiments to test whether participants could identify which images were genuine and which were computer-generated. Even when shown comparison photos or when participants were familiar with the faces depicted, their ability to spot fakes did not significantly improve.
Professor Jeremy Tree from Swansea University’s School of Psychology commented: “Studies have shown that face images of fictional people generated using AI are indistinguishable from real photographs. But for this research we went further by generating synthetic images of real people.
“The fact that everyday AI tools can do this not only raises urgent concerns about misinformation and trust in visual media but also the need for reliable detection methods as a matter of urgency.”
In one experiment involving participants from the US, Canada, UK, Australia, and New Zealand, subjects viewed both authentic and AI-generated faces and attempted to determine which was which. Many mistook synthetic faces for real ones. In another experiment focused on celebrities such as Paul Rudd and Olivia Wilde, participants again struggled to tell genuine photos apart from AI-created versions.
The findings raise questions about how easily manipulated images could be used to mislead the public—for example, by creating fake endorsements by celebrities or political figures.
Professor Tree said: “This study shows that AI can create synthetic images of both new and known faces that most people can’t tell apart from real photos. Familiarity with a face or having reference images didn’t help much in spotting the fakes, that is why we urgently need to find new ways to detect them.
“While automated systems may eventually outperform humans at this task, for now, it’s up to viewers to judge what’s real.”
The full research findings have been published in Cognitive Research: Principles and Implications.