A better understanding of how people judge the authenticity of images is necessary given the rise of AI-generated false content, particularly realistic images used in political disinformation. The current study examined the effects of emotional salience and aesthetic realism on how people view AI-generated disinformation images. Results from 292 participants in the UK revealed that realistic AI images were more frequently, but less confidently, regarded as authentic. The impact of emotional salience was minimal. User confidence and image realism were key factors in determining trust in AI detection tools. These observations can guide the creation of instruments and strategies to combat visual disinformation.

https://link.springer.com/article/10.1007/s00146-025-02416-5